{ "stats": { "run_id": "20261007T205849Z", "generated_at": "2026-10-07T20:59:10+00:00", "records": { "github_repos": 708, "hn": 53, "hn_hiring": 240, "arxiv": 147, "github_people": 243 }, "records_total": 1391, "repos_rejected": { "big company org": 33, "list/tutorial/course description": 77, "list/tutorial/course name": 84, "inflated star pattern (forks under 1% of stars)": 1, "archived": 1 }, "entities": 701, "multi_source_entities": 274, "enriched": 90, "agent_calls": 50, "models": { "claude-sonnet-5-5": 50 }, "input_tokens": 268430, "output_tokens": 24803, "cost_usd": 0.7849, "cost_cap_usd": 5.0, "duration_s": 59, "shown": 23, "claims_verified": 103, "claims_discarded": 10, "filtered_off_thesis": [ { "name": "Ping.gg", "url": "https://ping.gg", "score": 56.8 }, { "name": "Hypit.AI", "url": "https://hypit.ai", "score": 53.6 }, { "name": "LuwuDynamics", "url": "https://xgorobot.com", "score": 49.8 }, { "name": "OpenWAM", "url": "https://github.com/OpenWAM-Official", "score": 43.4 }, { "name": "Omacom", "url": "https://omarchy.org", "score": 41.0 }, { "name": "z2labs", "url": "https://z2labs.io", "score": 41.0 }, { "name": "BuzzPlay", "url": "https://buzzplay.app", "score": 39.9 }, { "name": "house-planner", "url": "https://homedesignsai.pro", "score": 39.6 }, { "name": "Autonomous", "url": "https://autonomous.ai", "score": 38.6 }, { "name": "godot-pty", "url": "https://github.com/godot-pty", "score": 38.6 }, { "name": "agent-me", "url": "https://jz-ai-chat.com", "score": 38.5 }, { "name": "HydraTerms", "url": "https://hydraterms.com", "score": 38.0 }, { "name": "RonanRx", "url": "https://ronanrx.com", "score": 38.0 }, { "name": "status-trio", "url": "https://lingai.net", "score": 37.7 }, { "name": "object-permanence", "url": "https://object-permanence.world", "score": 37.1 }, { "name": "Voight", "url": "https://voight.xyz", "score": 36.6 }, { "name": "superplothq", "url": "https://superplot.dev", "score": 35.7 }, { "name": "Real-time Solar System with 526k asteroids and all tracked s", "url": "https://bl2.net", "score": 35.5 }, { "name": "matlab-simulink-energy-lab", "url": "https://khanlab.co.technology", "score": 34.5 }, { "name": "Alibaba AMAP CV Lab", "url": "https://github.com/amap-cvlab", "score": 32.7 }, { "name": "cozmo-esp32", "url": "https://instagram.com", "score": 32.5 }, { "name": "LynnReal-AI", "url": "https://github.com/LynnReal-AI", "score": 32.4 }, { "name": "GuitarPedalSTM32", "url": "https://guitarml.com", "score": 31.8 }, { "name": "NFC Energy-Harvesting PCB Business Card", "url": "https://wilsonharper.net", "score": 31.3 }, { "name": "Against All Odds (SEEDS, PasRISCV)", "url": "https://againstallodds.games", "score": 31.2 } ], "not_enough_evidence": [ { "name": "Hebbian Robotics", "url": "https://hebbianrobotics.com", "score": 57.4 }, { "name": "ESPARGOS", "url": "https://espargos.net", "score": 48.1 } ] }, "cards": [ { "rank": 1, "id": "dom:magnitude.dev", "name": "Magnitude", "area": "AI infrastructure", "url": "https://magnitude.dev", "is_company": true, "score": 74.5, "components": { "velocity": { "points": 21.8, "cap": 25, "why": "55.2 stars/day on magnitudedev/magnitude (6486 stars in 118 days)" }, "launch": { "points": 13.9, "cap": 15, "why": "194 points on HN (launch hn)" }, "recency": { "points": 10.0, "cap": 10, "why": "newest signal 7 days old" }, "founder": { "points": 5.0, "cap": 20, "why": "YC (anerli)" }, "corroboration": { "points": 10.0, "cap": 15, "why": "3 source types: GitHub, Hacker News, own domain" }, "formation": { "points": 10, "cap": 10, "why": "org-owned (magnitudedev), own domain magnitude.dev, company launch on HN" }, "thesis": { "points": 3.8, "cap": 5, "why": "AI infrastructure, Semiconductors" } }, "claims": { "what": [ { "text": "They build an open source inference engine in Rust that compiles and tunes GPU kernels on the user's own device to run open models for local agents.", "quote": "We built it in Rust, including a custom GPU kernel runtime and autotuner.", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49911995", "source_kind": "Hacker News", "source_label": "HN, 194 points" } ], "why_now": [ { "text": "Existing engines trade off between datacenter batching, broad compatibility, or narrow hardware focus, and none are built for long, concurrent local agent sessions.", "quote": "Plus none of them are designed for running agents locally. Sessions are long, several often run at once, and you still want to use your computer for other things.", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49911995", "source_kind": "Hacker News", "source_label": "HN, 194 points" } ], "founder": [], "open_question": [] }, "discarded": [ { "field": "why_now", "text": "They report a decode speedup over llama.cpp on Apple Silicon with a 35B mixture-of-experts model.", "quote": "92% faster decode (30 tok/s \u2192 57 tok/s)", "source_id": "S7", "reason": "not in cited source: apple, silicon" }, { "field": "founder", "text": "The two founders previously built an open source browser agent that reached thousands of GitHub stars.", "quote": "previously built an open source browser agent to 4k+ GH stars and 100k+ downloads", "source_id": "S7", "reason": "not in cited source: github" }, { "field": "open_question", "text": "Benchmarks cover one model on two machines, so how well do the speedups hold across other model families, quantizations, and hardware such as AMD and CPU-only?", "quote": "Benchmarked against llama.cpp with Qwen 3.6 35B A3B (4 bit), 64k context, no speculative decoding:", "source_id": "S7", "reason": "not in cited source: amd, cpu-only" } ], "links": [ { "label": "magnitudedev/magnitude", "url": "https://github.com/magnitudedev/magnitude" }, { "label": "HN 194 pts", "url": "https://news.ycombinator.com/item?id=49911995" } ], "first_seen": "2026-10-07", "stars": 6486, "velocity": 55.2, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/magnitudedev/magnitude", "label": "magnitudedev/magnitude", "fetched_at": "2026-10-07T20:40:12+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/magnitudedev/magnitude/blob/main/README.md", "label": "magnitudedev/magnitude README", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/magnitudedev", "label": "magnitudedev", "fetched_at": "2026-10-07T20:40:22+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/anerli", "label": "anerli", "fetched_at": "2026-10-07T20:40:27+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/thrgreenwald", "label": "thrgreenwald", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S6", "kind": "GitHub profile", "url": "https://github.com/ambikeesshh", "label": "ambikeesshh", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S7", "kind": "Hacker News", "url": "https://news.ycombinator.com/item?id=49911995", "label": "HN, 194 points", "fetched_at": "2026-10-07T20:58:52+00:00" }, { "id": "S8", "kind": "Homepage", "url": "https://magnitude.dev", "label": "magnitude.dev", "fetched_at": "2026-10-07T20:40:28+00:00" } ] }, { "rank": 2, "id": "dom:experientiallabs.ai", "name": "Experiential Labs", "area": "AI infrastructure", "url": "https://experientiallabs.ai", "is_company": true, "score": 71.3, "components": { "velocity": { "points": 24.5, "cap": 25, "why": "90.99 stars/day on experientiallabs/experiential (9547 stars in 105 days)" }, "launch": { "points": 14.2, "cap": 15, "why": "222 points on HN (show hn)" }, "recency": { "points": 6.4, "cap": 10, "why": "newest signal 41 days old" }, "founder": { "points": 8.0, "cap": 20, "why": "PhD (kfallah), top lab or school (kfallah)" }, "corroboration": { "points": 10.0, "cap": 15, "why": "3 source types: GitHub, Hacker News, own domain" }, "formation": { "points": 7.0, "cap": 10, "why": "org-owned (experientiallabs), own domain experientiallabs.ai" }, "thesis": { "points": 1.2, "cap": 5, "why": "AI infrastructure" } }, "claims": { "what": [ { "text": "They build an open source gateway that routes requests to hosted, BYOK and local models through one OpenAI-compatible API.", "quote": "Use hosted, BYOK, and local models through one OpenAI-compatible API.", "source_id": "S2", "source_url": "https://github.com/experientiallabs/experiential/blob/main/README.md", "source_kind": "README", "source_label": "experientiallabs/experiential README" } ], "why_now": [ { "text": "They use production traces to fit a router and train a custom model, aiming to cut cost versus calling frontier models directly.", "quote": "given standardized OTel traces, we mine representative real tasks, use text world models to simulate rollouts for various models, apply an LLM judge", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49471407", "source_kind": "Hacker News", "source_label": "HN, 222 points" }, { "text": "They report a distilled 9B model that cut cost on a computer use agent task set while running faster than a frontier model.", "quote": "A 9B model distilled from a teacher working 1,000 tasks on Mac VMs", "source_id": "S8", "source_url": "https://experientiallabs.ai", "source_kind": "Homepage", "source_label": "experientiallabs.ai" } ], "founder": [], "open_question": [ { "text": "How much of the intelligence layer's cost savings holds up on customer workloads beyond the two published case studies, given the router is a nearest neighbor classifier on prompt embeddings?", "quote": "Usually this can map out a better pareto curve on cost/quality than just calling single models but it\u2019s not perfect.", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49471407", "source_kind": "Hacker News", "source_label": "HN, 222 points" } ] }, "discarded": [ { "field": "founder", "text": "One listed contributor is a Georgia Tech ML PhD student with an electrical and computer engineering degree from Ohio State.", "quote": "PhD ML@gatech \\ ECE Ohio State '18", "source_id": "S5", "reason": "quote not found in cited source" } ], "links": [ { "label": "experientiallabs/experiential", "url": "https://github.com/experientiallabs/experiential" }, { "label": "HN 222 pts", "url": "https://news.ycombinator.com/item?id=49471407" } ], "first_seen": "2026-10-07", "stars": 9547, "velocity": 90.99, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/experientiallabs/experiential", "label": "experientiallabs/experiential", "fetched_at": "2026-10-07T20:40:11+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/experientiallabs/experiential/blob/main/README.md", "label": "experientiallabs/experiential README", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/experientiallabs", "label": "experientiallabs", "fetched_at": "2026-10-07T20:40:19+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/SilenNaihin", "label": "SilenNaihin", "fetched_at": "2026-10-07T20:40:27+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/kfallah", "label": "kfallah", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S6", "kind": "GitHub profile", "url": "https://github.com/Adityakk9031", "label": "Adityakk9031", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S7", "kind": "Hacker News", "url": "https://news.ycombinator.com/item?id=49471407", "label": "HN, 222 points", "fetched_at": "2026-10-07T20:58:50+00:00" }, { "id": "S8", "kind": "Homepage", "url": "https://experientiallabs.ai", "label": "experientiallabs.ai", "fetched_at": "2026-10-07T20:40:28+00:00" } ] }, { "rank": 3, "id": "dom:reefinfra.ai", "name": "Human Agent Society", "area": "AI agents and models", "url": "https://reefinfra.ai", "is_company": true, "score": 61.9, "components": { "velocity": { "points": 25.0, "cap": 25, "why": "198.07 stars/day on Human-Agent-Society/reef (7488 stars in 38 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 6.9, "cap": 10, "why": "newest signal 38 days old" }, "founder": { "points": 5.0, "cap": 20, "why": "PhD (Benjamin-eecs)" }, "corroboration": { "points": 10.0, "cap": 15, "why": "3 source types: GitHub, arXiv, own domain" }, "formation": { "points": 10.0, "cap": 10, "why": "org-owned (Human-Agent-Society), org created 206 days ago, own domain reefinfra.ai" }, "thesis": { "points": 5.0, "cap": 5, "why": "AI infrastructure, AI agents and models" } }, "claims": { "what": [ { "text": "They build Reef, open-source infrastructure that serves agent requests, collects feedback, and updates model weights or the agent harness.", "quote": "It connects agent inference, feedback, learning, and versioned delivery. Use it to train model weights with Slime and SGLang", "source_id": "S2", "source_url": "https://github.com/Human-Agent-Society/reef/blob/main/README.md", "source_kind": "README", "source_label": "Human-Agent-Society/reef README" } ], "why_now": [ { "text": "Reef targets agents that learn from live traffic, a gap between inference engines and RL training frameworks.", "quote": "The first inference-native infrastructure for agents to learn from live traffic and continually update their weights and harness", "source_id": "S7", "source_url": "https://reefinfra.ai", "source_kind": "Homepage", "source_label": "reefinfra.ai" }, { "text": "The repo has drawn early open-source traction with many contributors within about a month of launch.", "quote": "227 commits and 242 merged pull requests in 31 days.", "source_id": "S7", "source_url": "https://reefinfra.ai", "source_kind": "Homepage", "source_label": "reefinfra.ai" } ], "founder": [ { "text": "One listed person is an incoming Stanford NLP PhD student working on self-improving agents.", "quote": "Incoming CS PhD @stanfordnlp Building continually self-improving foundation model agents", "source_id": "S5", "source_url": "https://github.com/Benjamin-eecs", "source_kind": "GitHub profile", "source_label": "Benjamin-eecs" } ], "open_question": [ { "text": "What is the business model, given that Reef is Apache-2.0 open source and what would customers pay for?", "quote": "Reef is open source under Apache-2.0", "source_id": "S7", "source_url": "https://reefinfra.ai", "source_kind": "Homepage", "source_label": "reefinfra.ai" } ] }, "discarded": [], "links": [ { "label": "Human-Agent-Society/reef", "url": "https://github.com/Human-Agent-Society/reef" }, { "label": "arXiv 2610.06269v1", "url": "https://arxiv.org/abs/2610.06269v1" } ], "first_seen": "2026-10-07", "stars": 7488, "velocity": 198.07, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/Human-Agent-Society/reef", "label": "Human-Agent-Society/reef", "fetched_at": "2026-10-07T20:39:52+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/Human-Agent-Society/reef/blob/main/README.md", "label": "Human-Agent-Society/reef README", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/Human-Agent-Society", "label": "Human-Agent-Society", "fetched_at": "2026-10-07T20:40:14+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/BobbyZhouZijian", "label": "BobbyZhouZijian", "fetched_at": "2026-10-07T20:40:27+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/Benjamin-eecs", "label": "Benjamin-eecs", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S6", "kind": "arXiv", "url": "https://arxiv.org/abs/2610.06269v1", "label": "arXiv 2610.06269v1", "fetched_at": "2026-10-07T20:33:09+00:00" }, { "id": "S7", "kind": "Homepage", "url": "https://reefinfra.ai", "label": "reefinfra.ai", "fetched_at": "2026-10-07T20:40:28+00:00" } ] }, { "rank": 4, "id": "dom:inco.ai", "name": "Inco AI", "area": "AI infrastructure", "url": "https://inco.ai", "is_company": true, "score": 57.1, "components": { "velocity": { "points": 22.8, "cap": 25, "why": "66.7 stars/day on incoai/splash (1270 stars in 19 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 9.3, "cap": 10, "why": "newest signal 19 days old" }, "founder": { "points": 5.0, "cap": 20, "why": "PhD (jianc99)" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: GitHub, own domain" }, "formation": { "points": 10.0, "cap": 10, "why": "org-owned (incoai), org created 155 days ago, own domain inco.ai" }, "thesis": { "points": 5.0, "cap": 5, "why": "AI infrastructure, Semiconductors" } }, "claims": { "what": [ { "text": "They build Splash, a local inference engine for Apple silicon that serves coding agents and OpenAI or Anthropic compatible apps on one Mac.", "quote": "Splash runs coding agents and OpenAI or Anthropic compatible applications on one Mac.", "source_id": "S2", "source_url": "https://github.com/incoai/splash/blob/main/README.md", "source_kind": "README", "source_label": "incoai/splash README" } ], "why_now": [ { "text": "Splash reports decode speeds of 2.5 to 5.3 times llama.cpp on the same weights, which makes agent workloads on a laptop more practical.", "quote": "That is 2.5-3.2\u00d7 as fast on the 35B and 4.5-5.3\u00d7 on the 27B", "source_id": "S2", "source_url": "https://github.com/incoai/splash/blob/main/README.md", "source_kind": "README", "source_label": "incoai/splash README" }, { "text": "The company positions its inference work around agentic use, and the repo already has over a thousand stars.", "quote": "Inco AI, inference, reimagined for the agentic era.", "source_id": "S7", "source_url": "https://inco.ai", "source_kind": "Homepage", "source_label": "inco.ai" } ], "founder": [ { "text": "One GitHub profile linked to the project belongs to a UCSD Ph.D. student.", "quote": "Ph.D. student at UCSD", "source_id": "S4", "source_url": "https://github.com/jianc99", "source_kind": "GitHub profile", "source_label": "jianc99" } ], "open_question": [ { "text": "Who are the founders, and how do they plan to turn an Apache-2.0 local engine into a business?", "quote": "Apache-2.0. GGUF kernels include MIT-licensed material from llama.cpp; see third-party notices.", "source_id": "S2", "source_url": "https://github.com/incoai/splash/blob/main/README.md", "source_kind": "README", "source_label": "incoai/splash README" } ] }, "discarded": [], "links": [ { "label": "incoai/splash", "url": "https://github.com/incoai/splash" } ], "first_seen": "2026-10-07", "stars": 1270, "velocity": 66.7, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/incoai/splash", "label": "incoai/splash", "fetched_at": "2026-10-07T20:38:57+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/incoai/splash/blob/main/README.md", "label": "incoai/splash README", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/incoai", "label": "incoai", "fetched_at": "2026-10-07T20:40:20+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/jianc99", "label": "jianc99", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/linson007", "label": "linson007", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S6", "kind": "GitHub profile", "url": "https://github.com/obekt", "label": "obekt", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S7", "kind": "Homepage", "url": "https://inco.ai", "label": "inco.ai", "fetched_at": "2026-10-07T20:40:28+00:00" } ] }, { "rank": 5, "id": "dom:kingkong.tech", "name": "KingKong Robotics", "area": "Robotics", "url": "https://kingkong.tech", "is_company": true, "score": 55.0, "components": { "velocity": { "points": 25.0, "cap": 25, "why": "138.99 stars/day on KingKongRobotics/jumper (1338 stars in 10 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 10.0, "cap": 10, "why": "newest signal 10 days old" }, "founder": { "points": 0.0, "cap": 20, "why": "no founder signal in bios" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: GitHub, own domain" }, "formation": { "points": 10.0, "cap": 10, "why": "org-owned (KingKongRobotics), org created 11 days ago, own domain kingkong.tech" }, "thesis": { "points": 5.0, "cap": 5, "why": "Robotics" } }, "claims": { "what": [ { "text": "They build a 22-DoF open-source crab robot called Jumper, with an AI toolkit for skins, motion training and scenes.", "quote": "Jumper is a 22-DoF crab robot. View hardware \u2192", "source_id": "S2", "source_url": "https://github.com/KingKongRobotics/jumper/blob/main/README.md", "source_kind": "README", "source_label": "KingKongRobotics/jumper README" } ], "why_now": [ { "text": "Their motion training stack uses open simulation and RL tooling, so gait policies can be trained in simulation and exported to the robot.", "quote": "Training builds on mjlab, rsl_rl, MuJoCo and MuJoCo Warp.", "source_id": "S2", "source_url": "https://github.com/KingKongRobotics/jumper/blob/main/README.md", "source_kind": "README", "source_label": "KingKongRobotics/jumper README" }, { "text": "The company also sells its own robot components, including encoders, which suggests vertical control of small-robot hardware.", "quote": "\u91d1\u94a2\u79d1\u6280\u4e3a\u673a\u5668\u4eba\u8d4b\u4e88\u9ad8\u654f\u611f\u7684\u795e\u7ecf\u7cfb\u7edf\u7f16\u7801\u5668\u4ea7\u54c1\uff0c\u6253\u9020\u5f3a\u800c\u654f\u611f\u7684\u89e6\u89c9\u8235\u673a\uff0c\u6784\u5efa\u5404\u79cd\u5c0f\u578b\u5316\u673a\u5668\u4eba\u4ea7\u54c1", "source_id": "S5", "source_url": "https://kingkong.tech", "source_kind": "Homepage", "source_label": "kingkong.tech" } ], "founder": [ { "text": "A GitHub profile lists Zheng Liang as affiliated with KingKong Robotics.", "quote": "Zheng Liang KingKong Robotics https://kingkong.tech", "source_id": "S4", "source_url": "https://github.com/zhiyishou", "source_kind": "GitHub profile", "source_label": "zhiyishou" } ], "open_question": [] }, "discarded": [ { "field": "open_question", "text": "Is the main business the encoder and servo components or the Jumper crab robot, and who is paying for each today?", "quote": "\u52a0\u901f\u8ba9\u5c0f\u673a\u5668\u4eba\u8d70\u5165\u5bb6\u5ead\uff0c\u611f\u77e5\u6e29\u5ea6\u3001\u5e26\u6765\u6b22\u4e50", "source_id": "S5", "reason": "not in cited source: jumper" } ], "links": [ { "label": "KingKongRobotics/jumper", "url": "https://github.com/KingKongRobotics/jumper" } ], "first_seen": "2026-10-07", "stars": 1338, "velocity": 138.99, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/KingKongRobotics/jumper", "label": "KingKongRobotics/jumper", "fetched_at": "2026-10-07T20:37:42+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/KingKongRobotics/jumper/blob/main/README.md", "label": "KingKongRobotics/jumper README", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/KingKongRobotics", "label": "KingKongRobotics", "fetched_at": "2026-10-07T20:40:15+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/zhiyishou", "label": "zhiyishou", "fetched_at": "2026-10-07T20:40:28+00:00" }, { "id": "S5", "kind": "Homepage", "url": "https://kingkong.tech", "label": "kingkong.tech", "fetched_at": "2026-10-07T20:40:29+00:00" } ] }, { "rank": 6, "id": "dom:datologyai.com", "name": "Datology AI", "area": "AI infrastructure", "url": "https://datologyai.com", "is_company": true, "score": 54.4, "components": { "velocity": { "points": 20.6, "cap": 25, "why": "44 stars/day on datologyai/zephon (44 stars in 1 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 10.0, "cap": 10, "why": "newest signal 1 days old" }, "founder": { "points": 8.0, "cap": 20, "why": "PhD (MaxiBoether), top lab or school (MaxiBoether)" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: GitHub, own domain" }, "formation": { "points": 7.0, "cap": 10, "why": "org-owned (datologyai), own domain datologyai.com" }, "thesis": { "points": 3.8, "cap": 5, "why": "AI infrastructure, AI agents and models" } }, "claims": { "what": [ { "text": "Datology sells a data curation pipeline that cleans, curates, synthesizes and mixes training data for teams building their own models.", "quote": "Datology turns your data into high-quality training data for mid-training on open models or pre-training your own foundation models, all deployed on your private cloud.", "source_id": "S7", "source_url": "https://datologyai.com", "source_kind": "Homepage", "source_label": "datologyai.com" } ], "why_now": [ { "text": "They claim a ~10 person team trained a frontier-class open-weights MoE on 20T tokens curated by Datology, which supports the case that data quality substitutes for budget.", "quote": "A frontier-class open weights MoE built without a frontier budget or research team. ~10 people, 20T tokens curated by Datology.", "source_id": "S7", "source_url": "https://datologyai.com", "source_kind": "Homepage", "source_label": "datologyai.com" }, { "text": "Their open-source loader Zephon addresses jobs that get preempted and resume on fewer GPUs, which can silently change the data a run sees.", "quote": "if a job gets preempted and comes back on fewer GPUs (which happens more often than any of us would like), many loaders will either refuse to resume", "source_id": "S2", "source_url": "https://github.com/datologyai/zephon/blob/main/README.md", "source_kind": "README", "source_label": "datologyai/zephon README" } ], "founder": [ { "text": "Josh Wills, creator of dbt-duckdb, is a member of technical staff at Datology.", "quote": "Member of Technical Staff @ DatologyAI, creator of dbt-duckdb", "source_id": "S5", "source_url": "https://github.com/jwills", "source_kind": "GitHub profile", "source_label": "jwills" } ], "open_question": [ { "text": "How much of the frontier open-weights model's result is attributable to Datology curation versus the model team's own work, and is there a controlled comparison?", "quote": "Competitive with models trained by 170-person teams that raised $2B.", "source_id": "S7", "source_url": "https://datologyai.com", "source_kind": "Homepage", "source_label": "datologyai.com" } ] }, "discarded": [], "links": [ { "label": "datologyai/zephon", "url": "https://github.com/datologyai/zephon" } ], "first_seen": "2026-10-07", "stars": 44, "velocity": 44.0, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/datologyai/zephon", "label": "datologyai/zephon", "fetched_at": "2026-10-07T20:39:43+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/datologyai/zephon/blob/main/README.md", "label": "datologyai/zephon README", "fetched_at": "2026-10-07T20:40:29+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/datologyai", "label": "datologyai", "fetched_at": "2026-10-07T20:40:19+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/MaxiBoether", "label": "MaxiBoether", "fetched_at": "2026-10-07T20:40:29+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/jwills", "label": "jwills", "fetched_at": "2026-10-07T20:40:29+00:00" }, { "id": "S6", "kind": "GitHub profile", "url": "https://github.com/danielzayas", "label": "danielzayas", "fetched_at": "2026-10-07T20:40:29+00:00" }, { "id": "S7", "kind": "Homepage", "url": "https://datologyai.com", "label": "datologyai.com", "fetched_at": "2026-10-07T20:40:30+00:00" } ] }, { "rank": 7, "id": "dom:mirros.ai", "name": "MirroS", "area": "AI agents and models", "url": "https://mirros.ai", "is_company": true, "score": 54.1, "components": { "velocity": { "points": 13.7, "cap": 25, "why": "11.55 stars/day on MirroS-Lab/Code-as-World (475 stars in 41 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 6.4, "cap": 10, "why": "newest signal 41 days old" }, "founder": { "points": 14.0, "cap": 20, "why": "PhD (Manchery), founder (Manchery), top lab or school (Manchery)" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: GitHub, own domain" }, "formation": { "points": 10.0, "cap": 10, "why": "org-owned (MirroS-Lab), org created 60 days ago, own domain mirros.ai" }, "thesis": { "points": 5.0, "cap": 5, "why": "AI agents and models" } }, "claims": { "what": [ { "text": "They build vision-language models that turn video observations into executable simulation code for physical reasoning.", "quote": "Code-as-World introduces code as executable representations for the physical world and an agentic process for discovering them through iterative simulation and verification.", "source_id": "S2", "source_url": "https://github.com/MirroS-Lab/Code-as-World/blob/main/README.md", "source_kind": "README", "source_label": "MirroS-Lab/Code-as-World README" } ], "why_now": [ { "text": "They also ship an agent-based benchmark for world models, which gives them an evaluation tool alongside their models.", "quote": "HarnessEval is an agentified evaluation pipeline that brings the harness paradigm from the LLM ecosystem to world model benchmarking.", "source_id": "S4", "source_url": "https://github.com/MirroS-Lab/HarnessEval-W/blob/main/README.md", "source_kind": "README", "source_label": "MirroS-Lab/HarnessEval-W README" }, { "text": "Their stated approach turns abundant raw observations into physical supervision data for training models.", "quote": "it turns raw abundant observations into reusable physical data: explicit states, dynamics, and mechanisms that capture not only what was seen", "source_id": "S2", "source_url": "https://github.com/MirroS-Lab/Code-as-World/blob/main/README.md", "source_kind": "README", "source_label": "MirroS-Lab/Code-as-World README" } ], "founder": [ { "text": "Fangfu Liu is a Tsinghua PhD student in electronic engineering and the CEO of MirroS-Lab.", "quote": "I am currently a PhD student majoring in Electronic Engineering at Tsinghua University, and also the CEO of @MirroS-Lab", "source_id": "S7", "source_url": "https://github.com/liuff19", "source_kind": "GitHub profile", "source_label": "liuff19" } ], "open_question": [ { "text": "What is the path from open-source research releases to a product with paying customers, and who would buy it?", "quote": "MirroS builds physical intelligence beyond the known world.", "source_id": "S9", "source_url": "https://mirros.ai", "source_kind": "Homepage", "source_label": "mirros.ai" } ] }, "discarded": [ { "field": "founder", "text": "Jialong Wu describes himself as a co-founder of MirroS-Lab and a Tsinghua PhD student working on world models.", "quote": "Co-founder @MirroS-Lab, Ph.D. student @thuml, in pursuit of world models", "source_id": "S6", "reason": "not in cited source: phd" } ], "links": [ { "label": "MirroS-Lab/Code-as-World", "url": "https://github.com/MirroS-Lab/Code-as-World" }, { "label": "MirroS-Lab/HarnessEval-W", "url": "https://github.com/MirroS-Lab/HarnessEval-W" } ], "first_seen": "2026-10-07", "stars": 475, "velocity": 11.55, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/MirroS-Lab/Code-as-World", "label": "MirroS-Lab/Code-as-World", "fetched_at": "2026-10-07T20:39:50+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/MirroS-Lab/Code-as-World/blob/main/README.md", "label": "MirroS-Lab/Code-as-World README", "fetched_at": "2026-10-07T20:40:31+00:00" }, { "id": "S3", "kind": "GitHub repo", "url": "https://github.com/MirroS-Lab/HarnessEval-W", "label": "MirroS-Lab/HarnessEval-W", "fetched_at": "2026-10-07T20:39:49+00:00" }, { "id": "S4", "kind": "README", "url": "https://github.com/MirroS-Lab/HarnessEval-W/blob/main/README.md", "label": "MirroS-Lab/HarnessEval-W README", "fetched_at": "2026-10-07T20:40:31+00:00" }, { "id": "S5", "kind": "GitHub org", "url": "https://github.com/MirroS-Lab", "label": "MirroS-Lab", "fetched_at": "2026-10-07T20:40:15+00:00" }, { "id": "S6", "kind": "GitHub profile", "url": "https://github.com/Manchery", "label": "Manchery", "fetched_at": "2026-10-07T20:40:30+00:00" }, { "id": "S7", "kind": "GitHub profile", "url": "https://github.com/liuff19", "label": "liuff19", "fetched_at": "2026-10-07T20:40:31+00:00" }, { "id": "S8", "kind": "GitHub profile", "url": "https://github.com/chen-wl20", "label": "chen-wl20", "fetched_at": "2026-10-07T20:40:31+00:00" }, { "id": "S9", "kind": "Homepage", "url": "https://mirros.ai", "label": "mirros.ai", "fetched_at": "2026-10-07T20:40:32+00:00" } ] }, { "rank": 8, "id": "dom:general-instinct.com", "name": "General Instinct", "area": "Robotics", "url": "https://general-instinct.com", "is_company": true, "score": 53.3, "components": { "velocity": { "points": 4.9, "cap": 25, "why": "1.46 stars/day on General-Instinct/InstinctFlash (194 stars in 133 days)" }, "launch": { "points": 8.8, "cap": 15, "why": "27 points on HN (show hn)" }, "recency": { "points": 9.8, "cap": 10, "why": "newest signal 15 days old" }, "founder": { "points": 6.0, "cap": 20, "why": "founder (william-Dic)" }, "corroboration": { "points": 10.0, "cap": 15, "why": "3 source types: GitHub, Hacker News, own domain" }, "formation": { "points": 10.0, "cap": 10, "why": "org-owned (General-Instinct), org created 133 days ago, own domain general-instinct.com" }, "thesis": { "points": 3.8, "cap": 5, "why": "Robotics, AI infrastructure" } }, "claims": { "what": [ { "text": "They build InstinctFlash, an open-source serving runtime that speeds up robotics models on Jetson Thor and RTX GPUs.", "quote": "We just released InstinctFlash, a high-performance serving framework for robotics models. It\u2019s licensed under AGPL-3.0.", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49802789", "source_kind": "Hacker News", "source_label": "HN, 27 points" } ], "why_now": [ { "text": "Robot policies are improving but inference is often too slow for the control loop teams want to run.", "quote": "the models were getting much better, but inference was often way too slow for the control loop we actually wanted", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49802789", "source_kind": "Hacker News", "source_label": "HN, 27 points" }, { "text": "With few-step distillation, they report a 33.78x speedup on LingBot-VA on Jetson Thor.", "quote": "up to 33.78x for LingBot-VA when we combine those runtime optimizations with a distilled few-step diffusion scheduler", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49802789", "source_kind": "Hacker News", "source_label": "HN, 27 points" } ], "founder": [ { "text": "Guanming Wang is a co-founder of General Instinct and lists it as a YC P26 company.", "quote": "Co-Founder of General Instinct (YC P26)", "source_id": "S5", "source_url": "https://github.com/william-Dic", "source_kind": "GitHub profile", "source_label": "william-Dic" } ], "open_question": [ { "text": "Is there a durable business beyond the AGPL open-source runtime, and who pays for it, given the claimed use at Samsung and Siemens?", "quote": "Teams at Samsung, Siemens, and other robotics startups have used InstinctFlash for model acceleration on VLAs, WAMs, and diffusion-based world models.", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49802789", "source_kind": "Hacker News", "source_label": "HN, 27 points" } ] }, "discarded": [], "links": [ { "label": "General-Instinct/InstinctFlash", "url": "https://github.com/General-Instinct/InstinctFlash" }, { "label": "HN 27 pts", "url": "https://news.ycombinator.com/item?id=49802789" } ], "first_seen": "2026-10-07", "stars": 194, "velocity": 1.46, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/General-Instinct/InstinctFlash", "label": "General-Instinct/InstinctFlash", "fetched_at": "2026-10-07T20:40:11+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/General-Instinct/InstinctFlash/blob/main/README.md", "label": "General-Instinct/InstinctFlash README", "fetched_at": "2026-10-07T20:40:29+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/General-Instinct", "label": "General-Instinct", "fetched_at": "2026-10-07T20:40:14+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/LH-and-FPGA", "label": "LH-and-FPGA", "fetched_at": "2026-10-07T20:40:29+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/william-Dic", "label": "william-Dic", "fetched_at": "2026-10-07T20:40:29+00:00" }, { "id": "S6", "kind": "GitHub profile", "url": "https://github.com/n26modi", "label": "n26modi", "fetched_at": "2026-10-07T20:40:29+00:00" }, { "id": "S7", "kind": "Hacker News", "url": "https://news.ycombinator.com/item?id=49802789", "label": "HN, 27 points", "fetched_at": "2026-10-07T20:58:50+00:00" }, { "id": "S8", "kind": "Homepage", "url": "https://general-instinct.com", "label": "general-instinct.com", "fetched_at": "2026-10-07T20:40:30+00:00" } ] }, { "rank": 9, "id": "dom:voygr.tech", "name": "VOYGR", "area": "AI agents and models", "url": "https://voygr.tech", "is_company": true, "score": 51.6, "components": { "velocity": { "points": 2.0, "cap": 25, "why": "0.46 stars/day on voygr-tech/placecall (35 stars in 76 days)" }, "launch": { "points": 7.4, "cap": 15, "why": "16 points on HN (show hn)" }, "recency": { "points": 10.0, "cap": 10, "why": "newest signal 13 days old" }, "founder": { "points": 11.0, "cap": 20, "why": "founder (borzelo), YC (borzelo)" }, "corroboration": { "points": 10.0, "cap": 15, "why": "3 source types: GitHub, Hacker News, own domain" }, "formation": { "points": 10.0, "cap": 10, "why": "org-owned (voygr-tech), org created 274 days ago, own domain voygr.tech" }, "thesis": { "points": 1.2, "cap": 5, "why": "AI agents and models" } }, "claims": { "what": [ { "text": "VOYGR sells an API that lets AI agents phone US businesses, navigate menus and hold, and return an outcome, recording and transcript.", "quote": "it navigates phone menus, waits for you on hold, and returns a structured response with outcome, recording and transcript.", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49831231", "source_kind": "Hacker News", "source_label": "HN, 16 points" } ], "why_now": [ { "text": "Agents can find a business online but still cannot get a human on the phone, which is how many businesses are reached.", "quote": "Assistants can\u2019t do much in the real world where the phone number is still a major communication interface.", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49831231", "source_kind": "Hacker News", "source_label": "HN, 16 points" }, { "text": "The team says existing voice agent tools need assembly, and nothing works out of the box for agent builders.", "quote": "There are a bunch of tools to create your own voice agents, but nothing that can work simply out of the box.", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49831231", "source_kind": "Hacker News", "source_label": "HN, 16 points" } ], "founder": [ { "text": "The founders previously worked on Google Maps and Search.", "quote": "We are both from Google (Maps and Search) and even there, the merchants and venues don\u2019t keep this info updated.", "source_id": "S7", "source_url": "https://news.ycombinator.com/item?id=49831231", "source_kind": "Hacker News", "source_label": "HN, 16 points" } ], "open_question": [ { "text": "Can PlaceCall hold a high rate of verified outcomes at scale, and what does each successful call cost versus what customers pay?", "quote": "Pay only for calls that deliver a successful outcome", "source_id": "S8", "source_url": "https://voygr.tech", "source_kind": "Homepage", "source_label": "voygr.tech" } ] }, "discarded": [], "links": [ { "label": "voygr-tech/placecall", "url": "https://github.com/voygr-tech/placecall" }, { "label": "HN 16 pts", "url": "https://news.ycombinator.com/item?id=49831231" } ], "first_seen": "2026-10-07", "stars": 35, "velocity": 0.46, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/voygr-tech/placecall", "label": "voygr-tech/placecall", "fetched_at": "2026-10-07T20:42:42+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/voygr-tech/placecall/blob/main/README.md", "label": "voygr-tech/placecall README", "fetched_at": "2026-10-07T20:42:46+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/voygr-tech", "label": "voygr-tech", "fetched_at": "2026-10-07T20:42:43+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/ibukeev-voygr", "label": "ibukeev-voygr", "fetched_at": "2026-10-07T20:42:45+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/ibukeev", "label": "ibukeev", "fetched_at": "2026-10-07T20:42:45+00:00" }, { "id": "S6", "kind": "GitHub profile", "url": "https://github.com/borzelo", "label": "borzelo", "fetched_at": "2026-10-07T20:42:45+00:00" }, { "id": "S7", "kind": "Hacker News", "url": "https://news.ycombinator.com/item?id=49831231", "label": "HN, 16 points", "fetched_at": "2026-10-07T20:58:50+00:00" }, { "id": "S8", "kind": "Homepage", "url": "https://voygr.tech", "label": "voygr.tech", "fetched_at": "2026-10-07T20:42:46+00:00" } ] }, { "rank": 10, "id": "dom:norirobotics.com", "name": "Nori Robotics", "area": "Robotics", "url": "https://norirobotics.com", "is_company": true, "score": 50.5, "components": { "velocity": { "points": 0.6, "cap": 25, "why": "0.12 stars/day on Nori-Robotics/nori-sdk-py (6 stars in 48 days)" }, "launch": { "points": 14.0, "cap": 15, "why": "201 points on HN (launch hn)" }, "recency": { "points": 7.1, "cap": 10, "why": "newest signal 36 days old" }, "founder": { "points": 5.0, "cap": 20, "why": "YC (AntonioLi)" }, "corroboration": { "points": 10.0, "cap": 15, "why": "3 source types: GitHub, Hacker News, own domain" }, "formation": { "points": 10, "cap": 10, "why": "org-owned (Nori-Robotics), org created 122 days ago, own domain norirobotics.com, company launch on HN" }, "thesis": { "points": 3.8, "cap": 5, "why": "Robotics" } }, "claims": { "what": [ { "text": "They build and assemble in San Francisco a $1,688 bimanual mobile robot for robotics developers and researchers.", "quote": "We build a $1,688 bimanual mobile robot in San Francisco for robotics developers and researchers.", "source_id": "S6", "source_url": "https://news.ycombinator.com/item?id=49525153", "source_kind": "Hacker News", "source_label": "HN, 201 points" } ], "why_now": [ { "text": "Labs have few robots, which limits dataset collection, and Nori targets that with a low-cost platform.", "quote": "Most labs have one or two expensive robots, which makes it hard to collect large datasets, run long experiments, or test across several robots.", "source_id": "S6", "source_url": "https://news.ycombinator.com/item?id=49525153", "source_kind": "Hacker News", "source_label": "HN, 201 points" }, { "text": "They have already shipped a first robot and are building the next batch.", "quote": "We\u2019ve shipped our first robot and are building the next batch.", "source_id": "S6", "source_url": "https://news.ycombinator.com/item?id=49525153", "source_kind": "Hacker News", "source_label": "HN, 201 points" } ], "founder": [ { "text": "The founder began Nori while doing robotics research at Columbia, working on teaching robots from human demonstrations.", "quote": "I started working on Nori while doing robotics research at Columbia. I was teaching robots through human demonstrations", "source_id": "S6", "source_url": "https://news.ycombinator.com/item?id=49525153", "source_kind": "Hacker News", "source_label": "HN, 201 points" } ], "open_question": [ { "text": "With a Raspberry Pi 5 onboard and heavy policies run off-robot over LAN or WAN, can the robot support the learned manipulation tasks buyers expect, and how many units have customers actually bought and used?", "quote": "SLAM and safeties are run on board, heavier ACT and VLAs must be run from a computer via LAN or a server via WAN", "source_id": "S6", "source_url": "https://news.ycombinator.com/item?id=49525153", "source_kind": "Hacker News", "source_label": "HN, 201 points" } ] }, "discarded": [], "links": [ { "label": "Nori-Robotics/nori-sdk-py", "url": "https://github.com/Nori-Robotics/nori-sdk-py" }, { "label": "HN 201 pts", "url": "https://news.ycombinator.com/item?id=49525153" } ], "first_seen": "2026-10-07", "stars": 6, "velocity": 0.12, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/Nori-Robotics/nori-sdk-py", "label": "Nori-Robotics/nori-sdk-py", "fetched_at": "2026-10-07T20:42:43+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/Nori-Robotics/nori-sdk-py/blob/main/README.md", "label": "Nori-Robotics/nori-sdk-py README", "fetched_at": "2026-10-07T20:42:45+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/Nori-Robotics", "label": "Nori-Robotics", "fetched_at": "2026-10-07T20:42:43+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/ShanyingLiu", "label": "ShanyingLiu", "fetched_at": "2026-10-07T20:42:45+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/AlexanderRoller", "label": "AlexanderRoller", "fetched_at": "2026-10-07T20:42:45+00:00" }, { "id": "S6", "kind": "Hacker News", "url": "https://news.ycombinator.com/item?id=49525153", "label": "HN, 201 points", "fetched_at": "2026-10-07T20:58:52+00:00" }, { "id": "S7", "kind": "Homepage", "url": "https://norirobotics.com", "label": "norirobotics.com", "fetched_at": "2026-10-07T20:40:33+00:00" } ] }, { "rank": 11, "id": "dom:aetherlabs.ai", "name": "Aether AI (AetherLabs-AI)", "area": "Robotics", "url": "https://aetherlabs.ai", "is_company": true, "score": 48.0, "components": { "velocity": { "points": 18.0, "cap": 25, "why": "26.77 stars/day on AetherLabsAI/Video2World (53 stars in 2 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 10.0, "cap": 10, "why": "newest signal 2 days old" }, "founder": { "points": 0.0, "cap": 20, "why": "no founder signal in bios" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: GitHub, own domain" }, "formation": { "points": 10.0, "cap": 10, "why": "org-owned (AetherLabsAI), org created 28 days ago, own domain aetherlabs.ai" }, "thesis": { "points": 5.0, "cap": 5, "why": "Robotics, AI agents and models" } }, "claims": { "what": [ { "text": "They build causal world models, starting with a decision layer for robots between perception and control.", "quote": "Aether is building the decision brain for Physical AI, the intelligence layer between perception and control", "source_id": "S5", "source_url": "https://aetherlabs.ai", "source_kind": "Homepage", "source_label": "aetherlabs.ai" } ], "why_now": [ { "text": "They released a benchmark testing whether coding agents can turn embodied videos into working simulators, which targets the cost of scaling robot data.", "quote": "Building interactive simulators from real-world observations is a promising way to scale embodied data, but current pipelines still rely heavily on manual environment construction and calibration.", "source_id": "S2", "source_url": "https://github.com/AetherLabsAI/Video2World/blob/main/README.md", "source_kind": "README", "source_label": "AetherLabsAI/Video2World README" }, { "text": "They argue that current robot AI maps observations straight to actions and breaks when conditions change.", "quote": "Much of today's robotics AI still maps observations directly to actions. These systems can learn useful behaviors in familiar settings, but they become brittle", "source_id": "S5", "source_url": "https://aetherlabs.ai", "source_kind": "Homepage", "source_label": "aetherlabs.ai" } ], "founder": [ { "text": "The founding team describes itself as experts in causal discovery and causal AI, and the benchmark authors are affiliated with UC San Diego.", "quote": "Our founding team are leading experts in causal discovery, causal AI, causal foundation models, causal reinforcement learning, agentic systems, and foundation model training.", "source_id": "S5", "source_url": "https://aetherlabs.ai", "source_kind": "Homepage", "source_label": "aetherlabs.ai" } ], "open_question": [ { "text": "Beyond the benchmark, what working causal world model has the team built and shown to improve real robot performance over end-to-end policies?", "quote": "Aether is building a new class of AI systems that understand mechanisms, reason under intervention, and operate reliably in real-world systems.", "source_id": "S5", "source_url": "https://aetherlabs.ai", "source_kind": "Homepage", "source_label": "aetherlabs.ai" } ] }, "discarded": [], "links": [ { "label": "AetherLabsAI/Video2World", "url": "https://github.com/AetherLabsAI/Video2World" } ], "first_seen": "2026-10-07", "stars": 53, "velocity": 26.77, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/AetherLabsAI/Video2World", "label": "AetherLabsAI/Video2World", "fetched_at": "2026-10-07T20:37:46+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/AetherLabsAI/Video2World/blob/main/README.md", "label": "AetherLabsAI/Video2World README", "fetched_at": "2026-10-07T20:40:29+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/AetherLabsAI", "label": "AetherLabsAI", "fetched_at": "2026-10-07T20:40:13+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/tangjzh", "label": "tangjzh", "fetched_at": "2026-10-07T20:40:29+00:00" }, { "id": "S5", "kind": "Homepage", "url": "https://aetherlabs.ai", "label": "aetherlabs.ai", "fetched_at": "2026-10-07T20:40:29+00:00" } ] }, { "rank": 12, "id": "dom:nari-labs.com", "name": "Nari Labs", "area": "AI infrastructure", "url": "https://nari-labs.com", "is_company": true, "score": 48.0, "components": { "velocity": { "points": 7.8, "cap": 25, "why": "3.22 stars/day on nari-labs/nari-qwen3-tts (183 stars in 57 days)" }, "launch": { "points": 11.9, "cap": 15, "why": "92 points on HN (show hn)" }, "recency": { "points": 8.8, "cap": 10, "why": "newest signal 23 days old" }, "founder": { "points": 0.0, "cap": 20, "why": "no founder signal in bios" }, "corroboration": { "points": 10.0, "cap": 15, "why": "3 source types: GitHub, Hacker News, own domain" }, "formation": { "points": 7.0, "cap": 10, "why": "org-owned (nari-labs), own domain nari-labs.com" }, "thesis": { "points": 2.5, "cap": 5, "why": "AI infrastructure, AI agents and models" } }, "claims": { "what": [ { "text": "Nari Labs sells hosted speech-to-text and text-to-speech APIs built on its own inference engine for open models.", "quote": "Optimized multimodal model endpoints built to be fastest in production: starting with speech.", "source_id": "S6", "source_url": "https://nari-labs.com", "source_kind": "Homepage", "source_label": "nari-labs.com" } ], "why_now": [ { "text": "They argue closed speech models lead because open-model serving is slow, and that existing engines like vLLM and SGLang fit multimodal inference poorly.", "quote": "Existing systems such as vLLM / SGLang are not well suited for multimodal inference.", "source_id": "S5", "source_url": "https://news.ycombinator.com/item?id=49699267", "source_kind": "Hacker News", "source_label": "HN, 92 points" }, { "text": "Their open engine serves 10 requests per second at sub-50 ms p95 time-to-first-audio on one H100.", "quote": "It achieves 10 requests per second (RPS) and sub-50 ms p95 time-to-first-audio (TTFA) while maintaining real-time playback on a single NVIDIA H100 SXM.", "source_id": "S2", "source_url": "https://github.com/nari-labs/nari-qwen3-tts/blob/main/README.md", "source_kind": "README", "source_label": "nari-labs/nari-qwen3-tts README" } ], "founder": [ { "text": "The team previously built Dia, an open-source dialogue text-to-speech model.", "quote": "Last year, we built Dia, the first OSS text-to-speech model capable of doing natural dialogue.", "source_id": "S5", "source_url": "https://news.ycombinator.com/item?id=49699267", "source_kind": "Hacker News", "source_label": "HN, 92 points" } ], "open_question": [] }, "discarded": [ { "field": "open_question", "text": "Is the inference speedup a durable moat, or will vLLM-Omni and SGLang-Omni close the gap and leave Nari competing on price alone?", "quote": "We want to continue to push prices down to make speech technology a commodity", "source_id": "S5", "reason": "not in cited source: omni, sglang-omni" } ], "links": [ { "label": "nari-labs/nari-qwen3-tts", "url": "https://github.com/nari-labs/nari-qwen3-tts" }, { "label": "HN 92 pts", "url": "https://news.ycombinator.com/item?id=49699267" } ], "first_seen": "2026-10-07", "stars": 183, "velocity": 3.22, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/nari-labs/nari-qwen3-tts", "label": "nari-labs/nari-qwen3-tts", "fetched_at": "2026-10-07T20:42:42+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/nari-labs/nari-qwen3-tts/blob/main/README.md", "label": "nari-labs/nari-qwen3-tts README", "fetched_at": "2026-10-07T20:42:45+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/nari-labs", "label": "nari-labs", "fetched_at": "2026-10-07T20:42:43+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/narilabs", "label": "narilabs", "fetched_at": "2026-10-07T20:42:45+00:00" }, { "id": "S5", "kind": "Hacker News", "url": "https://news.ycombinator.com/item?id=49699267", "label": "HN, 92 points", "fetched_at": "2026-10-07T20:58:50+00:00" }, { "id": "S6", "kind": "Homepage", "url": "https://nari-labs.com", "label": "nari-labs.com", "fetched_at": "2026-10-07T20:42:46+00:00" } ] }, { "rank": 13, "id": "dom:holiday-robotics.com", "name": "Holiday Robotics", "area": "Robotics", "url": "https://holiday-robotics.com", "is_company": true, "score": 44.3, "components": { "velocity": { "points": 12.3, "cap": 25, "why": "8.68 stars/day on Holiday-Robot/FlashDexRetarget (30 stars in 4 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 10.0, "cap": 10, "why": "newest signal 3 days old" }, "founder": { "points": 5.0, "cap": 20, "why": "PhD (kyungminn)" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: GitHub, own domain" }, "formation": { "points": 7.0, "cap": 10, "why": "org-owned (Holiday-Robot), own domain holiday-robotics.com" }, "thesis": { "points": 5.0, "cap": 5, "why": "Robotics, AI agents and models" } }, "claims": { "what": [ { "text": "Holiday Robotics builds dexterous humanoids for industrial work.", "quote": "Holiday Robotics builds dexterous humanoids for real industrial work.", "source_id": "S9", "source_url": "https://holiday-robotics.com", "source_kind": "Homepage", "source_label": "holiday-robotics.com" } ], "why_now": [ { "text": "They released a method that retargets many human hand-object demonstrations to a robot hand with one jointly trained RL policy, which targets the data bottleneck in dexterous manipulation.", "quote": "retargets large collections of human hand-object demonstrations to a dexterous robot hand by training one RL policy jointly across all reference motions", "source_id": "S2", "source_url": "https://github.com/Holiday-Robot/FlashDexRetarget/blob/main/README.md", "source_kind": "README", "source_label": "Holiday-Robot/FlashDexRetarget README" }, { "text": "They also ship an off-policy RL implementation for high-dimensional robot control, benchmarked on locomotion, in-hand reposing and Franka manipulation tasks.", "quote": "12 tasks (velocity locomotion, dexterous in-hand reposing, Franka manipulation) trained for 50M environment steps at 1024 parallel environments", "source_id": "S4", "source_url": "https://github.com/Holiday-Robot/rsl_rl_flashsac/blob/main/README.md", "source_kind": "README", "source_label": "Holiday-Robot/rsl_rl_flashsac README" } ], "founder": [], "open_question": [ { "text": "What is the first deployed humanoid hand or customer use case, and how much of the sim RL work has transferred to real hardware?", "quote": "Holiday Robotics builds dexterous humanoids for real industrial work.", "source_id": "S9", "source_url": "https://holiday-robotics.com", "source_kind": "Homepage", "source_label": "holiday-robotics.com" } ] }, "discarded": [ { "field": "founder", "text": "Team members publish with KAIST AI researchers, and the corresponding author on the retargeting paper is at Holiday Robotics.", "quote": "Jaegul Choo 1\u2020 , Hojoon Lee 2\u2020 1 KAIST AI, 2 Holiday Robotics", "source_id": "S2", "reason": "quote not found in cited source" } ], "links": [ { "label": "Holiday-Robot/FlashDexRetarget", "url": "https://github.com/Holiday-Robot/FlashDexRetarget" }, { "label": "Holiday-Robot/rsl_rl_flashsac", "url": "https://github.com/Holiday-Robot/rsl_rl_flashsac" } ], "first_seen": "2026-10-07", "stars": 30, "velocity": 8.68, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/Holiday-Robot/FlashDexRetarget", "label": "Holiday-Robot/FlashDexRetarget", "fetched_at": "2026-10-07T20:39:52+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/Holiday-Robot/FlashDexRetarget/blob/main/README.md", "label": "Holiday-Robot/FlashDexRetarget README", "fetched_at": "2026-10-07T20:40:32+00:00" }, { "id": "S3", "kind": "GitHub repo", "url": "https://github.com/Holiday-Robot/rsl_rl_flashsac", "label": "Holiday-Robot/rsl_rl_flashsac", "fetched_at": "2026-10-07T20:39:52+00:00" }, { "id": "S4", "kind": "README", "url": "https://github.com/Holiday-Robot/rsl_rl_flashsac/blob/main/README.md", "label": "Holiday-Robot/rsl_rl_flashsac README", "fetched_at": "2026-10-07T20:40:33+00:00" }, { "id": "S5", "kind": "GitHub org", "url": "https://github.com/Holiday-Robot", "label": "Holiday-Robot", "fetched_at": "2026-10-07T20:40:14+00:00" }, { "id": "S6", "kind": "GitHub profile", "url": "https://github.com/kyungminn", "label": "kyungminn", "fetched_at": "2026-10-07T20:40:32+00:00" }, { "id": "S7", "kind": "GitHub profile", "url": "https://github.com/youngdolee-holiday", "label": "youngdolee-holiday", "fetched_at": "2026-10-07T20:40:33+00:00" }, { "id": "S8", "kind": "GitHub profile", "url": "https://github.com/anahrendra-holiday", "label": "anahrendra-holiday", "fetched_at": "2026-10-07T20:40:33+00:00" }, { "id": "S9", "kind": "Homepage", "url": "https://holiday-robotics.com", "label": "holiday-robotics.com", "fetched_at": "2026-10-07T20:40:33+00:00" } ] }, { "rank": 14, "id": "dom:qymis.tech", "name": "QymIs.Tech (QymCAD)", "area": "Manufacturing", "url": "https://qymis.tech", "is_company": false, "score": 40.6, "components": { "velocity": { "points": 14.4, "cap": 25, "why": "13.36 stars/day on QymIs-Tech/QymCAD (576 stars in 43 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 6.2, "cap": 10, "why": "newest signal 43 days old" }, "founder": { "points": 0.0, "cap": 20, "why": "no founder signal in bios" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: GitHub, own domain" }, "formation": { "points": 10.0, "cap": 10, "why": "org-owned (QymIs-Tech), org created 59 days ago, own domain qymis.tech" }, "thesis": { "points": 5.0, "cap": 5, "why": "Manufacturing, AI infrastructure" } }, "claims": { "what": [ { "text": "They build QymCAD, a desktop parametric CAD for mechanical parts and assemblies on the OpenCASCADE B-rep kernel.", "quote": "QymCAD is a desktop CAD for mechanical parts and assemblies: housings, brackets, mechanisms, printed and milled work.", "source_id": "S2", "source_url": "https://github.com/QymIs-Tech/QymCAD/blob/main/README.md", "source_kind": "README", "source_label": "QymIs-Tech/QymCAD README" } ], "why_now": [ { "text": "The tool is local-first with no cloud or subscription, and exchanges files with other CAD systems through STEP.", "quote": "Sketch \u2192 part \u2192 assembly. One program, one project file, no cloud and no subscription.", "source_id": "S2", "source_url": "https://github.com/QymIs-Tech/QymCAD/blob/main/README.md", "source_kind": "README", "source_label": "QymIs-Tech/QymCAD README" }, { "text": "It is a young build that is already usable for real parts but changes daily, with the CAM module not working yet.", "quote": "A development build. The program works and is fit for real parts, but it is updated daily.", "source_id": "S2", "source_url": "https://github.com/QymIs-Tech/QymCAD/blob/main/README.md", "source_kind": "README", "source_label": "QymIs-Tech/QymCAD README" } ], "founder": [ { "text": "The site describes the operation as a one-person design bureau run by a self-taught engineer.", "quote": "\u041a\u043e\u043d\u0441\u0442\u0440\u0443\u043a\u0442\u043e\u0440\u0441\u043a\u043e\u0435 \u0431\u044e\u0440\u043e \u0438 \u0437\u0430\u0432\u043e\u0434 \u0438\u0437 \u043e\u0434\u043d\u043e\u0433\u043e \u0447\u0435\u043b\u043e\u0432\u0435\u043a\u0430.", "source_id": "S6", "source_url": "https://qymis.tech", "source_kind": "Homepage", "source_label": "qymis.tech" }, { "text": "The README names Denis Kazachenkov as the author.", "quote": "QymIs Tech, qymis.tech. Author: Denis Kazachenkov.", "source_id": "S2", "source_url": "https://github.com/QymIs-Tech/QymCAD/blob/main/README.md", "source_kind": "README", "source_label": "QymIs-Tech/QymCAD README" } ], "open_question": [] }, "discarded": [ { "field": "open_question", "text": "Is QymCAD the company's focus, or a side project to a one-person hardware shop (headphones, netbook), and who would pay for it with an AGPL license?", "quote": "\u0420\u0430\u0437\u0440\u0430\u0431\u043e\u0442\u043a\u0430 \u0432\u0435\u0434\u0451\u0442\u0441\u044f \u0437\u0430 \u0441\u0432\u043e\u0439 \u0441\u0447\u0451\u0442. \u0415\u0441\u043b\u0438 \u0445\u043e\u0442\u0438\u0442\u0435 \u043f\u043e\u043c\u043e\u0447\u044c, \u0431\u0443\u0434\u0443 \u0431\u043b\u0430\u0433\u043e\u0434\u0430\u0440\u0435\u043d.", "source_id": "S6", "reason": "not in cited source: agpl" } ], "links": [ { "label": "QymIs-Tech/QymCAD", "url": "https://github.com/QymIs-Tech/QymCAD" } ], "first_seen": "2026-10-07", "stars": 576, "velocity": 13.36, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/QymIs-Tech/QymCAD", "label": "QymIs-Tech/QymCAD", "fetched_at": "2026-10-07T20:38:52+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/QymIs-Tech/QymCAD/blob/main/README.md", "label": "QymIs-Tech/QymCAD README", "fetched_at": "2026-10-07T20:40:31+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/QymIs-Tech", "label": "QymIs-Tech", "fetched_at": "2026-10-07T20:40:16+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/basson", "label": "basson", "fetched_at": "2026-10-07T20:40:30+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/styx", "label": "styx", "fetched_at": "2026-10-07T20:40:30+00:00" }, { "id": "S6", "kind": "Homepage", "url": "https://qymis.tech", "label": "qymis.tech", "fetched_at": "2026-10-07T20:40:31+00:00" } ] }, { "rank": 15, "id": "gh:agamrossen", "name": "VolAnti", "area": "Autonomy and defense", "url": "https://github.com/agamrossen/VolAnti", "is_company": true, "score": 39.6, "components": { "velocity": { "points": 14.2, "cap": 25, "why": "12.63 stars/day on agamrossen/VolAnti (407 stars in 32 days)" }, "launch": { "points": 7.4, "cap": 15, "why": "16 points on HN (show hn)" }, "recency": { "points": 8.0, "cap": 10, "why": "newest signal 29 days old" }, "founder": { "points": 0.0, "cap": 20, "why": "no founder signal in bios" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: GitHub, Hacker News" }, "formation": { "points": 0.0, "cap": 10, "why": "personal project" }, "thesis": { "points": 5.0, "cap": 5, "why": "Manufacturing, Autonomy and defense, Semiconductors" } }, "claims": { "what": [ { "text": "They build an open-source box with four microphones, an ESP32-S3 and a LoRa radio that detects drones by the sound of their propellers.", "quote": "A 91 mm box with four microphones, an ESP32-S3, an e-paper screen and a LoRa radio.", "source_id": "S2", "source_url": "https://github.com/agamrossen/VolAnti/blob/main/README.md", "source_kind": "README", "source_label": "agamrossen/VolAnti README" } ], "why_now": [ { "text": "Fibre-optic FPV drones emit no radio signal, so radio-based detection cannot see them.", "quote": "A fibre-optic FPV drone does not. It trails a thread of glass back to its operator, video comes down the fibre", "source_id": "S2", "source_url": "https://github.com/agamrossen/VolAnti/blob/main/README.md", "source_kind": "README", "source_label": "agamrossen/VolAnti README" }, { "text": "Field testing reached 104 m detection of a hovering rig with traffic and people talking nearby.", "quote": "Tier 4 fired at 104.2 m, 114 yards on the map, with people talking near the unit. Nothing fired on the cars.", "source_id": "S2", "source_url": "https://github.com/agamrossen/VolAnti/blob/main/README.md", "source_kind": "README", "source_label": "agamrossen/VolAnti README" } ], "founder": [ { "text": "The builder is an engineering student at the University of York, originally from Israel.", "quote": "I'm an engineering student at the University of York, originally from Israel, studying in England.", "source_id": "S3", "source_url": "https://news.ycombinator.com/item?id=49618955", "source_kind": "Hacker News", "source_label": "HN, 16 points" } ], "open_question": [ { "text": "How does detection range and false alarm rate hold up in real wind and noise at the civilian sites where the nine units are deployed, versus the single 104 m test?", "quote": "Range moves by an order of magnitude with wind. Still air and a quiet site should give 100 to 200 m.", "source_id": "S2", "source_url": "https://github.com/agamrossen/VolAnti/blob/main/README.md", "source_kind": "README", "source_label": "agamrossen/VolAnti README" } ] }, "discarded": [], "links": [ { "label": "agamrossen/VolAnti", "url": "https://github.com/agamrossen/VolAnti" }, { "label": "HN 16 pts", "url": "https://news.ycombinator.com/item?id=49618955" } ], "first_seen": "2026-10-07", "stars": 407, "velocity": 12.63, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/agamrossen/VolAnti", "label": "agamrossen/VolAnti", "fetched_at": "2026-10-07T20:38:53+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/agamrossen/VolAnti/blob/main/README.md", "label": "agamrossen/VolAnti README", "fetched_at": "2026-10-07T20:40:31+00:00" }, { "id": "S3", "kind": "Hacker News", "url": "https://news.ycombinator.com/item?id=49618955", "label": "HN, 16 points", "fetched_at": "2026-10-07T20:58:50+00:00" } ] }, { "rank": 16, "id": "gh:deeplethe", "name": "DeepLethe", "area": "AI infrastructure", "url": "https://github.com/deeplethe", "is_company": true, "score": 39.6, "components": { "velocity": { "points": 25.0, "cap": 25, "why": "133.28 stars/day on deeplethe/utopia (8138 stars in 61 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 3.8, "cap": 10, "why": "newest signal 61 days old" }, "founder": { "points": 0.0, "cap": 20, "why": "no founder signal in bios" }, "corroboration": { "points": 0.0, "cap": 15, "why": "1 source types: GitHub" }, "formation": { "points": 7.0, "cap": 10, "why": "org-owned (deeplethe), org created 152 days ago" }, "thesis": { "points": 3.8, "cap": 5, "why": "AI agents and models" } }, "claims": { "what": [ { "text": "They build Utopia, an open-source enterprise knowledge system that stores facts in a bitemporal graph with an ontology, running on one Rust binary and Postgres.", "quote": "One Rust binary and one Postgres. Full-text search is embedded in the binary, vectors go in pgvector, and the job queue is a table", "source_id": "S2", "source_url": "https://github.com/deeplethe/utopia/blob/main/README.md", "source_kind": "README", "source_label": "deeplethe/utopia README" } ], "why_now": [ { "text": "Agents need a knowledge base they can trust, and Utopia can be deployed offline so companies keep it on their own hardware.", "quote": "It deploys offline, so a company can stand up a knowledge foundation, a decision core its agents can trust, and a compliance audit trail on hardware it controls.", "source_id": "S2", "source_url": "https://github.com/deeplethe/utopia/blob/main/README.md", "source_kind": "README", "source_label": "deeplethe/utopia README" }, { "text": "Agents such as Claude Desktop and Cursor can read the knowledge base through an MCP server with scoped permissions.", "quote": "An MCP server for every knowledge base, so Claude Desktop, Cursor, Workbuddy and other agent frameworks can connect, with fine-grained permissions.", "source_id": "S2", "source_url": "https://github.com/deeplethe/utopia/blob/main/README.md", "source_kind": "README", "source_label": "deeplethe/utopia README" } ], "founder": [ { "text": "A GitHub profile lists WaylandYang as affiliated with DeepLethe, with the deeplethe.com domain.", "quote": "DeepLethe @deeplethe Apex Intelligence @apxin-ai Singapore / China deeplethe.com", "source_id": "S4", "source_url": "https://github.com/WaylandYang", "source_kind": "GitHub profile", "source_label": "WaylandYang" } ], "open_question": [ { "text": "Which paying enterprise customers are running Utopia in production, and how does the company plan to make money from an open-source core?", "quote": "The enterprise world model built by DeepLethe. It is the first open substrate for knowledge engineering that learns passively and governs itself.", "source_id": "S2", "source_url": "https://github.com/deeplethe/utopia/blob/main/README.md", "source_kind": "README", "source_label": "deeplethe/utopia README" } ] }, "discarded": [], "links": [ { "label": "deeplethe/utopia", "url": "https://github.com/deeplethe/utopia" } ], "first_seen": "2026-10-07", "stars": 8138, "velocity": 133.28, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/deeplethe/utopia", "label": "deeplethe/utopia", "fetched_at": "2026-10-07T20:39:49+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/deeplethe/utopia/blob/main/README.md", "label": "deeplethe/utopia README", "fetched_at": "2026-10-07T20:40:32+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/deeplethe", "label": "deeplethe", "fetched_at": "2026-10-07T20:40:19+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/WaylandYang", "label": "WaylandYang", "fetched_at": "2026-10-07T20:40:31+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/Maya-Kid", "label": "Maya-Kid", "fetched_at": "2026-10-07T20:40:32+00:00" }, { "id": "S6", "kind": "GitHub profile", "url": "https://github.com/rollroyces", "label": "rollroyces", "fetched_at": "2026-10-07T20:40:32+00:00" } ] }, { "rank": 17, "id": "dom:risklytics.ai", "name": "Risklytics", "area": "Autonomy and defense", "url": "https://risklytics.ai", "is_company": true, "score": 38.8, "components": { "velocity": { "points": 0.0, "cap": 25, "why": "no repo" }, "launch": { "points": 10.5, "cap": 15, "why": "54 points on HN (launch hn)" }, "recency": { "points": 6.3, "cap": 10, "why": "newest signal 42 days old" }, "founder": { "points": 5.0, "cap": 20, "why": "YC (AlexRisio)" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: Hacker News, own domain" }, "formation": { "points": 7.0, "cap": 10, "why": "own domain risklytics.ai, company launch on HN" }, "thesis": { "points": 5.0, "cap": 5, "why": "Autonomy and defense, Robotics" } }, "claims": { "what": [ { "text": "Risklytics is a licensed insurance brokerage that places coverage for companies building robots, drones, autonomous systems and satellites.", "quote": "we run an insurance brokerage for companies building robots, drones, autonomous systems and satellites.", "source_id": "S1", "source_url": "https://news.ycombinator.com/item?id=49451495", "source_kind": "Hacker News", "source_label": "HN, 54 points" } ], "why_now": [ { "text": "In January ISO published add-on clauses that void coverage for losses arising from AI, which hits hardware companies buying insurance.", "quote": "in January ISO published clauses that void coverage for losses arising out of AI in any part of a process.", "source_id": "S1", "source_url": "https://news.ycombinator.com/item?id=49451495", "source_kind": "Hacker News", "source_label": "HN, 54 points" }, { "text": "Most existing policies were written for operations with a person at the controls, which leaves autonomous systems poorly covered.", "quote": "most of the policies that would cover them were written for operations with a person at the controls.", "source_id": "S2", "source_url": "https://risklytics.ai", "source_kind": "Homepage", "source_label": "risklytics.ai" } ], "founder": [ { "text": "The two founders, Sam and Alex, are on leave from Harvard.", "quote": "we're Sam and Alex, founders of Risklytics (https://risklytics.ai). We're both on leave from Harvard", "source_id": "S1", "source_url": "https://news.ycombinator.com/item?id=49451495", "source_kind": "Hacker News", "source_label": "HN, 54 points" } ], "open_question": [ { "text": "Brokerage commissions are 10-20% of premium, so what is the path from a commission business to something with software-like margins and defensibility as incumbents adapt?", "quote": "the insurance company pays us a commission when you get covered, usually 10-20% of what you are charged", "source_id": "S1", "source_url": "https://news.ycombinator.com/item?id=49451495", "source_kind": "Hacker News", "source_label": "HN, 54 points" } ] }, "discarded": [], "links": [ { "label": "HN 54 pts", "url": "https://news.ycombinator.com/item?id=49451495" } ], "first_seen": "2026-10-07", "stars": null, "velocity": null, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "Hacker News", "url": "https://news.ycombinator.com/item?id=49451495", "label": "HN, 54 points", "fetched_at": "2026-10-07T20:58:52+00:00" }, { "id": "S2", "kind": "Homepage", "url": "https://risklytics.ai", "label": "risklytics.ai", "fetched_at": "2026-10-07T20:40:35+00:00" } ] }, { "rank": 18, "id": "dom:baltavista.com", "name": "fastroute", "area": "Semiconductors", "url": "https://baltavista.com", "is_company": false, "score": 35.9, "components": { "velocity": { "points": 12.9, "cap": 25, "why": "9.91 stars/day on parisxmas/fastroute (81 stars in 8 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 10.0, "cap": 10, "why": "newest signal 8 days old" }, "founder": { "points": 0.0, "cap": 20, "why": "no founder signal in bios" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: GitHub, own domain" }, "formation": { "points": 3.0, "cap": 10, "why": "own domain baltavista.com" }, "thesis": { "points": 5.0, "cap": 5, "why": "Semiconductors, Manufacturing" } }, "claims": { "what": [ { "text": "They build a Rust port of the Freerouting PCB autorouter that reads Specctra DSN files and works with KiCad.", "quote": "PCB autorouter for KiCad and Specctra DSN, a Rust port of the Freerouting PCB autorouter.", "source_id": "S2", "source_url": "https://github.com/parisxmas/fastroute/blob/main/README.md", "source_kind": "README", "source_label": "parisxmas/fastroute README" } ], "why_now": [ { "text": "It runs the same 20 benchmark boards in far less total time than the Java original, which matters for hardware design iteration.", "quote": "| Sum of 20 boards | 2102 s | 510 s | 420 s |", "source_id": "S2", "source_url": "https://github.com/parisxmas/fastroute/blob/main/README.md", "source_kind": "README", "source_label": "parisxmas/fastroute README" }, { "text": "It adds features Freerouting lacks that high-speed board design needs, such as length matching and differential pairs.", "quote": "Length matching (--tune): groups of nets (buses, pair skew) are matched with meanders after routing", "source_id": "S2", "source_url": "https://github.com/parisxmas/fastroute/blob/main/README.md", "source_kind": "README", "source_label": "parisxmas/fastroute README" } ], "founder": [], "open_question": [ { "text": "Is there a team or company behind this, and would they commercialize it or keep it as an open-source tool?", "quote": "Fast open-source PCB autorouter for KiCad and Specctra DSN", "source_id": "S1", "source_url": "https://github.com/parisxmas/fastroute", "source_kind": "GitHub repo", "source_label": "parisxmas/fastroute" } ] }, "discarded": [], "links": [ { "label": "parisxmas/fastroute", "url": "https://github.com/parisxmas/fastroute" } ], "first_seen": "2026-10-07", "stars": 81, "velocity": 9.91, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/parisxmas/fastroute", "label": "parisxmas/fastroute", "fetched_at": "2026-10-07T20:38:46+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/parisxmas/fastroute/blob/main/README.md", "label": "parisxmas/fastroute README", "fetched_at": "2026-10-07T20:40:34+00:00" }, { "id": "S3", "kind": "GitHub profile", "url": "https://github.com/parisxmas", "label": "parisxmas", "fetched_at": "2026-10-07T20:40:23+00:00" }, { "id": "S4", "kind": "Homepage", "url": "https://baltavista.com", "label": "baltavista.com", "fetched_at": "2026-10-07T20:40:35+00:00" } ] }, { "rank": 19, "id": "dom:copperhead.sh", "name": "copperhead", "area": "Semiconductors", "url": "https://copperhead.sh", "is_company": true, "score": 35.8, "components": { "velocity": { "points": 8.7, "cap": 25, "why": "4 stars/day on copperheadhq/copperhead (326 stars in 82 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 1.1, "cap": 10, "why": "newest signal 82 days old" }, "founder": { "points": 6.0, "cap": 20, "why": "founder (animesh-chouhan)" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: GitHub, own domain" }, "formation": { "points": 10.0, "cap": 10, "why": "org-owned (copperheadhq), org created 61 days ago, own domain copperhead.sh" }, "thesis": { "points": 5.0, "cap": 5, "why": "Semiconductors, Manufacturing, AI agents and models" } }, "claims": { "what": [ { "text": "They build an open-source AI agent that designs, documents and validates PCBs by editing existing KiCad repositories.", "quote": "An AI agent that designs, documents, and validates real PCBs from a prompt, working directly on existing KiCad repositories.", "source_id": "S2", "source_url": "https://github.com/copperheadhq/copperhead/blob/main/README.md", "source_kind": "README", "source_label": "copperheadhq/copperhead README" } ], "why_now": [ { "text": "Design decisions spread across schematic, BOM and docs drift apart, and the mismatch shows up at bring-up when a respin is costly.", "quote": "The inconsistency is found at bring-up, and a respin costs 5,000 to 50,000 dollars and six to eight weeks.", "source_id": "S7", "source_url": "https://copperhead.sh", "source_kind": "Homepage", "source_label": "copperhead.sh" }, { "text": "The agent verifies its own edits by running KiCad's ERC and DRC checks, which makes agent-driven hardware design checkable.", "quote": "verifies its own work by running kicad-cli ERC/DRC until the checks pass.", "source_id": "S2", "source_url": "https://github.com/copperheadhq/copperhead/blob/main/README.md", "source_kind": "README", "source_label": "copperheadhq/copperhead README" } ], "founder": [ { "text": "The founder is an IIT Kharagpur graduate who works in engineering at JPMorgan Chase.", "quote": "Founder @copperheadhq | Engineering @jpmorganchase | IIT Kharagpur", "source_id": "S4", "source_url": "https://github.com/animesh-chouhan", "source_kind": "GitHub profile", "source_label": "animesh-chouhan" } ], "open_question": [ { "text": "Beyond the open-source Open Telegraph ESP32 board, how many real boards from outside users have been fabricated and worked first time using copperhead's output?", "quote": "Open Telegraph: pocket-size ESP32-S3 Morse key, built end to end with this workflow. Every decision, every check, every file is public.", "source_id": "S7", "source_url": "https://copperhead.sh", "source_kind": "Homepage", "source_label": "copperhead.sh" } ] }, "discarded": [], "links": [ { "label": "copperheadhq/copperhead", "url": "https://github.com/copperheadhq/copperhead" } ], "first_seen": "2026-10-07", "stars": 326, "velocity": 4.0, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/copperheadhq/copperhead", "label": "copperheadhq/copperhead", "fetched_at": "2026-10-07T20:38:46+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/copperheadhq/copperhead/blob/main/README.md", "label": "copperheadhq/copperhead README", "fetched_at": "2026-10-07T20:40:39+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/copperheadhq", "label": "copperheadhq", "fetched_at": "2026-10-07T20:40:19+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/animesh-chouhan", "label": "animesh-chouhan", "fetched_at": "2026-10-07T20:40:39+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/Chirag6722", "label": "Chirag6722", "fetched_at": "2026-10-07T20:40:39+00:00" }, { "id": "S6", "kind": "GitHub profile", "url": "https://github.com/AniketR10", "label": "AniketR10", "fetched_at": "2026-10-07T20:40:39+00:00" }, { "id": "S7", "kind": "Homepage", "url": "https://copperhead.sh", "label": "copperhead.sh", "fetched_at": "2026-10-07T20:40:39+00:00" } ] }, { "rank": 20, "id": "gh:simplejev", "name": "SimpleJev", "area": "AI agents and models", "url": "https://github.com/SimpleJev", "is_company": false, "score": 34.8, "components": { "velocity": { "points": 9.1, "cap": 25, "why": "4.37 stars/day on SimpleJev/JevAny (66 stars in 15 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 9.9, "cap": 10, "why": "newest signal 15 days old" }, "founder": { "points": 5.0, "cap": 20, "why": "PhD (weitianxin)" }, "corroboration": { "points": 0.0, "cap": 15, "why": "1 source types: GitHub" }, "formation": { "points": 7.0, "cap": 10, "why": "org-owned (SimpleJev), org created 11 days ago" }, "thesis": { "points": 3.8, "cap": 5, "why": "AI agents and models" } }, "claims": { "what": [ { "text": "They build open infrastructure to train, evaluate and deploy small decision models that pick one option from a candidate list and return probabilities.", "quote": "One API takes the state, question and candidate options, then directly returns a choice and its probabilities.", "source_id": "S2", "source_url": "https://github.com/SimpleJev/JevAny/blob/main/README.md", "source_kind": "README", "source_label": "SimpleJev/JevAny README" } ], "why_now": [ { "text": "In agent loops, the model takes over repeated choices from the LLM, and in the WebShop example this cut LLM calls from 9 to 4.", "quote": "Actions fall from 9 to 5, LLM calls from 9 to 4, tokens from 38,852 to 14,256", "source_id": "S2", "source_url": "https://github.com/SimpleJev/JevAny/blob/main/README.md", "source_kind": "README", "source_label": "SimpleJev/JevAny README" }, { "text": "The stated use cases include robot action selection, which ties the work to robotics control stacks.", "quote": "route support tickets, select tools, or choose a robot's next action.", "source_id": "S2", "source_url": "https://github.com/SimpleJev/JevAny/blob/main/README.md", "source_kind": "README", "source_label": "SimpleJev/JevAny README" } ], "founder": [], "open_question": [ { "text": "Gains are uneven across tasks: WebArena showed no success gain and higher LLM calls. Which real workloads give a repeatable cost and reward win over LLM-only?", "quote": "WebArena (6 pairs) | 50% \u2192 50% | LLM calls +5.6%, tokens +28.2%, time \u22120.4%", "source_id": "S2", "source_url": "https://github.com/SimpleJev/JevAny/blob/main/README.md", "source_kind": "README", "source_label": "SimpleJev/JevAny README" } ] }, "discarded": [], "links": [ { "label": "SimpleJev/JevAny", "url": "https://github.com/SimpleJev/JevAny" } ], "first_seen": "2026-10-07", "stars": 66, "velocity": 4.37, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/SimpleJev/JevAny", "label": "SimpleJev/JevAny", "fetched_at": "2026-10-07T20:39:52+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/SimpleJev/JevAny/blob/main/README.md", "label": "SimpleJev/JevAny README", "fetched_at": "2026-10-07T20:40:40+00:00" }, { "id": "S3", "kind": "GitHub profile", "url": "https://github.com/weitianxin", "label": "weitianxin", "fetched_at": "2026-10-07T20:40:39+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/ZhiningLiu1998", "label": "ZhiningLiu1998", "fetched_at": "2026-10-07T20:40:40+00:00" }, { "id": "S5", "kind": "GitHub profile", "url": "https://github.com/YuanchenBei", "label": "YuanchenBei", "fetched_at": "2026-10-07T20:40:40+00:00" } ] }, { "rank": 21, "id": "gh:philolabs", "name": "PhiloLabs", "area": "AI agents and models", "url": "https://github.com/PhiloLabs", "is_company": false, "score": 34.6, "components": { "velocity": { "points": 14.9, "cap": 25, "why": "14.75 stars/day on PhiloLabs/fable51-worlds (518 stars in 35 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 7.2, "cap": 10, "why": "newest signal 35 days old" }, "founder": { "points": 3.0, "cap": 20, "why": "top lab or school (christine1729)" }, "corroboration": { "points": 0.0, "cap": 15, "why": "1 source types: GitHub" }, "formation": { "points": 7.0, "cap": 10, "why": "org-owned (PhiloLabs), org created 239 days ago" }, "thesis": { "points": 2.5, "cap": 5, "why": "Manufacturing" } }, "claims": { "what": [ { "text": "They build agent swarms that turn a text, image or video brief into a walkable Three.js world generated entirely by code.", "quote": "Claude Fable 5.1 agent swarms take a brief - a sentence, a photograph, a clip - then research the place, model it, render it, and check it.", "source_id": "S2", "source_url": "https://github.com/PhiloLabs/fable51-worlds/blob/main/README.md", "source_kind": "README", "source_label": "PhiloLabs/fable51-worlds README" } ], "why_now": [ { "text": "The roadmap targets worlds as training and evaluation environments for agents, where the generating code also supplies reward and verification.", "quote": "Worlds as training and evaluation environments, where the same code that builds the place also supplies the reward and the verification signal", "source_id": "S2", "source_url": "https://github.com/PhiloLabs/fable51-worlds/blob/main/README.md", "source_kind": "README", "source_label": "PhiloLabs/fable51-worlds README" }, { "text": "They plan simulation for robotics where ground truth comes from the geometry itself.", "quote": "Scenes that change over time, and simulation where the ground truth comes free with the geometry", "source_id": "S2", "source_url": "https://github.com/PhiloLabs/fable51-worlds/blob/main/README.md", "source_kind": "README", "source_label": "PhiloLabs/fable51-worlds README" } ], "founder": [], "open_question": [ { "text": "Which of the roadmap items, agent environments or robotics simulation, is the product, and is there a team and customer behind it beyond this demo repo?", "quote": "A world is the substrate. What stands on it is the point.", "source_id": "S2", "source_url": "https://github.com/PhiloLabs/fable51-worlds/blob/main/README.md", "source_kind": "README", "source_label": "PhiloLabs/fable51-worlds README" } ] }, "discarded": [], "links": [ { "label": "PhiloLabs/fable51-worlds", "url": "https://github.com/PhiloLabs/fable51-worlds" } ], "first_seen": "2026-10-07", "stars": 518, "velocity": 14.75, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/PhiloLabs/fable51-worlds", "label": "PhiloLabs/fable51-worlds", "fetched_at": "2026-10-07T20:38:55+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/PhiloLabs/fable51-worlds/blob/main/README.md", "label": "PhiloLabs/fable51-worlds README", "fetched_at": "2026-10-07T20:40:37+00:00" }, { "id": "S3", "kind": "GitHub profile", "url": "https://github.com/christine1729", "label": "christine1729", "fetched_at": "2026-10-07T20:40:37+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/VerizonCao", "label": "VerizonCao", "fetched_at": "2026-10-07T20:40:37+00:00" } ] }, { "rank": 22, "id": "dom:bomwiki.com", "name": "BOMWiki", "area": "Manufacturing", "url": "https://bomwiki.com", "is_company": true, "score": 34.4, "components": { "velocity": { "points": 12.0, "cap": 25, "why": "8.15 stars/day on BOMWiki/partmode (511 stars in 63 days)" }, "launch": { "points": 0.0, "cap": 15, "why": "no HN launch" }, "recency": { "points": 3.6, "cap": 10, "why": "newest signal 63 days old" }, "founder": { "points": 0.0, "cap": 20, "why": "no founder signal in bios" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: GitHub, own domain" }, "formation": { "points": 10.0, "cap": 10, "why": "org-owned (BOMWiki), org created 97 days ago, own domain bomwiki.com" }, "thesis": { "points": 3.8, "cap": 5, "why": "Manufacturing" } }, "claims": { "what": [ { "text": "They build PartMode, an open-source parametric mechanical CAD that runs in the browser on OpenCascade WebAssembly.", "quote": "PartMode is browser-based mechanical CAD built on OpenCascade WebAssembly, replicad, and three.js.", "source_id": "S2", "source_url": "https://github.com/BOMWiki/partmode/blob/main/README.md", "source_kind": "README", "source_label": "BOMWiki/partmode README" } ], "why_now": [ { "text": "They expose hosted MCP tools so outside agents can inspect, preview and commit CAD changes.", "quote": "Hosted MCP tools for capabilities, inspection, queries, detached previews, revision-safe commits, browser artifacts, and opt-in server-headless STEP workflows", "source_id": "S2", "source_url": "https://github.com/BOMWiki/partmode/blob/main/README.md", "source_kind": "README", "source_label": "BOMWiki/partmode README" } ], "founder": [], "open_question": [ { "text": "The README says there is no simulation, CAM or certified analysis, so what is the business model and which paying engineering users come first given the AGPL license?", "quote": "PartMode does not currently provide engineering simulation, certified analysis, automated design approval, CAM, native DWG authoring", "source_id": "S2", "source_url": "https://github.com/BOMWiki/partmode/blob/main/README.md", "source_kind": "README", "source_label": "BOMWiki/partmode README" } ] }, "discarded": [ { "field": "why_now", "text": "Agents get typed, permissioned operations against the same document model and kernel that people use, which targets AI-driven hardware design.", "quote": "Agents use typed, permissioned operations against the same schema and kernel; browser-approved agents can edit the open project.", "source_id": "S2", "reason": "not in cited source: ai-driven" } ], "links": [ { "label": "BOMWiki/partmode", "url": "https://github.com/BOMWiki/partmode" } ], "first_seen": "2026-10-07", "stars": 511, "velocity": 8.15, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "GitHub repo", "url": "https://github.com/BOMWiki/partmode", "label": "BOMWiki/partmode", "fetched_at": "2026-10-07T20:38:52+00:00" }, { "id": "S2", "kind": "README", "url": "https://github.com/BOMWiki/partmode/blob/main/README.md", "label": "BOMWiki/partmode README", "fetched_at": "2026-10-07T20:40:35+00:00" }, { "id": "S3", "kind": "GitHub org", "url": "https://github.com/BOMWiki", "label": "BOMWiki", "fetched_at": "2026-10-07T20:40:13+00:00" }, { "id": "S4", "kind": "GitHub profile", "url": "https://github.com/protosphinx", "label": "protosphinx", "fetched_at": "2026-10-07T20:40:35+00:00" } ] }, { "rank": 23, "id": "dom:mireye.ai", "name": "Mireye", "area": "AI infrastructure", "url": "https://mireye.ai", "is_company": true, "score": 33.2, "components": { "velocity": { "points": 0.0, "cap": 25, "why": "no repo" }, "launch": { "points": 8.9, "cap": 15, "why": "28 points on HN (launch hn)" }, "recency": { "points": 7.3, "cap": 10, "why": "newest signal 34 days old" }, "founder": { "points": 5.0, "cap": 20, "why": "YC (anshchokshi)" }, "corroboration": { "points": 5.0, "cap": 15, "why": "2 source types: Hacker News, own domain" }, "formation": { "points": 7.0, "cap": 10, "why": "own domain mireye.ai, company launch on HN" }, "thesis": { "points": 0.0, "cap": 5, "why": "no thesis keyword" } }, "claims": { "what": [ { "text": "Mireye sells one API and MCP server that gives AI agents data, enrichment, tools and signals for any US location.", "quote": "I'm building the infrastructure AI agents use to make decisions about physical places: data, enrichment, tools, and signals for any US location, behind one API and MCP server.", "source_id": "S1", "source_url": "https://news.ycombinator.com/item?id=49552616", "source_kind": "Hacker News", "source_label": "HN, 28 points" } ], "why_now": [ { "text": "Frontier models hallucinate on questions about specific places, which blocked an insurer's underwriting agents.", "quote": "Frontier models keep hallucinating when asked a specific question about a specific place.", "source_id": "S1", "source_url": "https://news.ycombinator.com/item?id=49552616", "source_kind": "Hacker News", "source_label": "HN, 28 points" }, { "text": "Use cases already include data center and energy siting, which sit inside the infrastructure thesis.", "quote": "Screen a data center at 39.0438, -77.4874: slope, FEMA flood zone, nearest 345 kV line, substation voltage and status, and the county's interconnection queue.", "source_id": "S2", "source_url": "https://mireye.ai", "source_kind": "Homepage", "source_label": "mireye.ai" } ], "founder": [ { "text": "The founder was building construction agents before Mireye and ran into the missing physical context firsthand.", "quote": "Before Mireye I was building construction agents and hit this wall myself: my agent could reason about anything online but knew nothing about the ground under it.", "source_id": "S1", "source_url": "https://news.ycombinator.com/item?id=49552616", "source_kind": "Hacker News", "source_label": "HN, 28 points" } ], "open_question": [ { "text": "Usage is spread across 311 of 317 fields with no dominant use case, so which customer segment pays enough to anchor the business?", "quote": "311 of the 317 fields in the catalog get queried and no use case dominates.", "source_id": "S1", "source_url": "https://news.ycombinator.com/item?id=49552616", "source_kind": "Hacker News", "source_label": "HN, 28 points" } ] }, "discarded": [], "links": [ { "label": "HN 28 pts", "url": "https://news.ycombinator.com/item?id=49552616" } ], "first_seen": "2026-10-07", "stars": null, "velocity": null, "model": "claude-sonnet-5-5", "error": null, "sources": [ { "id": "S1", "kind": "Hacker News", "url": "https://news.ycombinator.com/item?id=49552616", "label": "HN, 28 points", "fetched_at": "2026-10-07T20:58:52+00:00" }, { "id": "S2", "kind": "Homepage", "url": "https://mireye.ai", "label": "mireye.ai", "fetched_at": "2026-10-07T20:42:45+00:00" } ] } ] }