Published text

Structured research records

1. devin won the frame — the whole industry now sells 'coding agents' and ships agent-native ides; cognition's sweep (models, agent, ide, enterprise contracts) positions it as the independent agent lab while the foundation-model companies treat coding as a feature. · cognition-agent-era-arguments

   Original location · source-1bbad291f76f701d0539 · source-ab68c048f622c5487cd3 · source-64ac502a1459aa6bdd73 · source-b2840ab1745c1bf7e839

   Original structured fields

   {
     "claim": "devin won the frame — the whole industry now sells 'coding agents' and ships agent-native ides; cognition's sweep (models, agent, ide, enterprise contracts) positions it as the independent agent lab while the foundation-model companies treat coding as a feature.",
     "id": "arg-frame-won",
     "side": "for",
     "sourceIds": [
       "src-ci-cognition-seriesd",
       "src-ci-openai-codex",
       "src-ci-google-jules",
       "src-ci-cognition-devin-desktop"
     ]
   }

2. the wedge is the mundane — migrations, legacy updates, security swarms, 'drudgery work': goldman's deployment and the enterprise partnerships describe a business automating the work engineers avoid, not the work they do. that is enough to be very large. · cognition-agent-era-arguments

   Original location · source-27be554657a19b7483e8 · source-dd9b0f8f0929d484fb44 · source-2e3148dc6befba8e5d15 · source-52d6624601297ebd17a9

   Original structured fields

   {
     "claim": "the wedge is the mundane — migrations, legacy updates, security swarms, 'drudgery work': goldman's deployment and the enterprise partnerships describe a business automating the work engineers avoid, not the work they do. that is enough to be very large.",
     "id": "arg-mundane-wedge",
     "side": "for",
     "sourceIds": [
       "src-ci-cnbc-goldman",
       "src-ci-headsup-migrations",
       "src-ci-cognition-gov",
       "src-hp-techzine-msft"
     ]
   }

3. the capability claims were always ahead of the product — the debunk was right about the demo, answer.ai was right about the reliability, and the 13.86% was a subset result marketed as a headline. the company's response — price cuts, supervision features, benchmark disclaimers — is an admission the launch oversold. · cognition-agent-era-arguments

   Original location · source-b0dafab463859ad5de28 · source-3e2dace97d27f013de2c · source-bd86c98a6793c69ef8c4 · source-034c7082737b3e1133af · source-0db751b75021fab92d04

   Original structured fields

   {
     "claim": "the capability claims were always ahead of the product — the debunk was right about the demo, answer.ai was right about the reliability, and the 13.86% was a subset result marketed as a headline. the company's response — price cuts, supervision features, benchmark disclaimers — is an admission the launch oversold.",
     "id": "arg-oversold",
     "side": "against",
     "sourceIds": [
       "src-ci-iob-debunk-video",
       "src-ci-answerai-devin",
       "src-ci-register-devin-bad",
       "src-ci-plainai-devin2",
       "src-ci-cognition-swe15"
     ]
   }

4. the moat is contested — 'everyone builds the coding agent': openai has codex, anthropic has claude code, google has jules plus windsurf's founders, cursor has the mindshare, and the openhands community replicates the surface area for free. independence is a strategy or a euphemism for 'no platform underneath.' · cognition-agent-era-arguments

   Original location · source-ab68c048f622c5487cd3 · source-75343f5d5daaea74a9ea · source-64ac502a1459aa6bdd73 · source-b87d95c58f8c4efe90a7 · source-96df22d52de2b697c293 · source-5d96dd0653e8ac0fc447

   Original structured fields

   {
     "claim": "the moat is contested — 'everyone builds the coding agent': openai has codex, anthropic has claude code, google has jules plus windsurf's founders, cursor has the mindshare, and the openhands community replicates the surface area for free. independence is a strategy or a euphemism for 'no platform underneath.'",
     "id": "arg-moat-contested",
     "side": "against",
     "sourceIds": [
       "src-ci-openai-codex",
       "src-ci-aicost-claudecode",
       "src-ci-google-jules",
       "src-hp-reuters-google-windsurf",
       "src-ci-opendevin-repo",
       "src-ci-bnt-cursor-share"
     ]
   }

5. the labor math is real but earlier than advertised — amodei's '90%' and cognition's internal '89–95%' describe the same trend from opposite sides of the cap table; the honest version is that agents absorb tickets faster than headcount shrinks, and the surplus shows up as throughput before it shows up as layoffs. · cognition-agent-era-arguments

   Original location · source-67737b0b0ce0e8a02d33 · source-ebeb343faa4710da4e12 · source-f23722382305071bb0f1 · source-1bbad291f76f701d0539 · source-9c8b8da0c4acd715460c

   Original structured fields

   {
     "claim": "the labor math is real but earlier than advertised — amodei's '90%' and cognition's internal '89–95%' describe the same trend from opposite sides of the cap table; the honest version is that agents absorb tickets faster than headcount shrinks, and the surplus shows up as throughput before it shows up as layoffs.",
     "id": "arg-labor-math",
     "side": "nuance",
     "sourceIds": [
       "src-ci-amodei-90",
       "src-ci-redwood-amodei",
       "src-ci-bi-devin-jobs",
       "src-ci-cognition-seriesd",
       "src-ci-fortune-goldman"
     ]
   }

6. the talent-machine question — the ioi pipeline supplied the founders and the mythos; whether it supplies the moat depends on whether competitive-programming skill transfers to agent systems, or whether it mostly transferred to storytelling. the windsurf bench being 'five or six people' cuts both ways. · cognition-agent-era-arguments

   Original location · source-6d3eaa02c5959695b1f4 · source-915cc5f7006791fc9a01 · source-b174a05dfca3de702d20 · source-9000b86d1e91e6cdee86

   Original structured fields

   {
     "claim": "the talent-machine question — the ioi pipeline supplied the founders and the mythos; whether it supplies the moat depends on whether competitive-programming skill transfers to agent systems, or whether it mostly transferred to storytelling. the windsurf bench being 'five or six people' cuts both ways.",
     "id": "arg-talent-machine",
     "side": "nuance",
     "sourceIds": [
       "src-hp-cognition-about",
       "src-hp-colossus-wu-tapes",
       "src-ci-jeffwang-neoshow",
       "src-hp-bloomberg-gold-medalist-coders"
     ]
   }

7. automatic programming · cognition-idea-lineage

   Original location · source-866b7c02ca55d3a1d84c · source-ccae308cc63e68ee2466

   Original structured fields

   {
     "contribution": "the dream itself — automatic programming goes back to the 1950s; the wikipedia lineage runs through program synthesis, end-user programming, and every 'natural language to code' promise. devin's 'first ai software engineer' claim inherits seventy years of the same promise, at the moment the models could nearly keep it.",
     "era": "1950s–",
     "id": "lin-automatic-programming",
     "name": "automatic programming",
     "sourceIds": [
       "src-ci-wikipedia-autoprog",
       "src-ci-cacm-end-of-programming"
     ]
   }

8. codex and the copilot era · cognition-idea-lineage

   Original location · source-eef0e6c4854564bf3f68 · source-60f035c084f29a27c33d

   Original structured fields

   {
     "contribution": "openai's original codex (2021) — the descendant of gpt-3 that powered github copilot and established the assistive paradigm: autocomplete, pair programming, a human still driving. the thing devin was positioned against from its first sentence.",
     "era": "2021",
     "id": "lin-codex-copilot",
     "name": "codex and the copilot era",
     "sourceIds": [
       "src-ci-openai-codex-2021",
       "src-ci-copilot-launch"
     ]
   }

9. swe-bench and its audits · cognition-idea-lineage

   Original location · source-521b4186c88ec2132123 · source-7fb1ba94cb65b0e117ca · source-d0e3c86f1a860f403712 · source-1ecd5bcad1a9647b2911

   Original structured fields

   {
     "contribution": "swe-bench (2023) — princeton's real-github-issue benchmark turned 'can an llm fix a bug' into a number. cognition exploited its asymmetries (the 25% subset, the unassisted framing); the audit literature (swe-bench+, verified) then eroded the ruler itself.",
     "era": "2023–",
     "id": "lin-swebench",
     "name": "swe-bench and its audits",
     "sourceIds": [
       "src-ci-swebench-paper",
       "src-ci-swebench-plus",
       "src-ci-swebench-verified",
       "src-hp-cognition-swebench-report"
     ]
   }

10. the agent convergence · cognition-idea-lineage

    Original location · source-e36f16e72ae999b85b10 · source-c992c96cdd103473ee96

    Original structured fields

    {
      "contribution": "the agent scaffolding lineage — tool use, planning loops, and long-horizon task decomposition converging in 2023–24; the arxiv 'converged' argument is that agentic coding systems landed on roughly the same architecture at roughly the same time.",
      "era": "2023–2024",
      "id": "lin-agent-convergence",
      "name": "the agent convergence",
      "sourceIds": [
        "src-ci-arxiv-converged",
        "src-ci-earlyterms-agents"
      ]
    }

11. the ioi pipeline · cognition-idea-lineage

    Original location · source-6d3eaa02c5959695b1f4 · source-915cc5f7006791fc9a01 · source-9000b86d1e91e6cdee86 · source-c15321fb46b7ed3bf5b8

    Original structured fields

    {
      "contribution": "the competitive-programming pipeline — ioi/icpc culture as recruiting and method: cognition's founders plus neal wu, korotkevich, and andrew he; peer companies scale, perplexity, and decagon came from the same circuit. the contest style — bounded problems, adversarial test cases, speed — shaped what 'a software engineer' meant to the people building one.",
      "era": "2000s–2023",
      "id": "lin-ioi-pipeline",
      "name": "the ioi pipeline",
      "sourceIds": [
        "src-hp-cognition-about",
        "src-hp-colossus-wu-tapes",
        "src-hp-bloomberg-gold-medalist-coders",
        "src-hp-wiki-korotkevich"
      ]
    }

12. from copilot to coding agent · cognition-idea-lineage

    Original location · source-24b7f0d0ba5fb7d9ce77 · source-ab68c048f622c5487cd3 · source-64ac502a1459aa6bdd73 · source-75343f5d5daaea74a9ea · source-1097c6dfa60fa364a20c

    Original structured fields

    {
      "contribution": "the copilot→agent vocabulary shift — 'ai that helps you code' to 'ai that does the ticket.' devin's launch forced the reframe; by 2025–26 the whole field (codex, jules, copilot coding agent, claude code, swe-1) had adopted the agent frame devin introduced.",
      "era": "2024–",
      "id": "lin-vocab-shift",
      "name": "from copilot to coding agent",
      "sourceIds": [
        "src-hp-cognition-introducing-devin",
        "src-ci-openai-codex",
        "src-ci-google-jules",
        "src-ci-aicost-claudecode",
        "src-ci-tc-swe1"
      ]
    }

13. the migration wedge · cognition-idea-lineage

    Original location · source-52d6624601297ebd17a9 · source-dd9b0f8f0929d484fb44 · source-27be554657a19b7483e8 · source-9c8b8da0c4acd715460c

    Original structured fields

    {
      "contribution": "the enterprise-migration wedge — the boring work nobody wants: code modernization, legacy upgrades, large-repo context. microsoft's azure partnership aimed devin at it in 2024; headsup's migrations writeup and goldman's 'drudgery work' framing are the idea finding its actual market.",
      "era": "2024–",
      "id": "lin-migration-wedge",
      "name": "the migration wedge",
      "sourceIds": [
        "src-hp-techzine-msft",
        "src-ci-headsup-migrations",
        "src-ci-cnbc-goldman",
        "src-ci-fortune-goldman"
      ]
    }

14. does 'independent agent lab' survive the foundation-model squeeze — can a company that rides claude, gpt, and its own swe models out-innovate the labs it depends on? · cognition-idea-open-questions

    Original location

    Original structured fields

    {
      "text": "does 'independent agent lab' survive the foundation-model squeeze — can a company that rides claude, gpt, and its own swe models out-innovate the labs it depends on?"
    }

15. what does an agent-hour actually cost — the acu pricing, the $20 plan, and the guarantee's estimator all describe the same opaque unit; when does someone outside cognition get to check the math? · cognition-idea-open-questions

    Original location

    Original structured fields

    {
      "text": "what does an agent-hour actually cost — the acu pricing, the $20 plan, and the guarantee's estimator all describe the same opaque unit; when does someone outside cognition get to check the math?"
    }

16. does the 13.86%→benchmark-decay arc repeat — will swe-2's self-reported numbers meet a swe-bench-pro audit, and does cognition keep a benchmark it can report? · cognition-idea-open-questions

    Original location

    Original structured fields

    {
      "text": "does the 13.86%→benchmark-decay arc repeat — will swe-2's self-reported numbers meet a swe-bench-pro audit, and does cognition keep a benchmark it can report?"
    }

17. was windsurf a rescue or a raid — the 'one boat' memo promised 100% participation; does the retained team's output at devin desktop vindicate the six-day memo? · cognition-idea-open-questions

    Original location

    Original structured fields

    {
      "text": "was windsurf a rescue or a raid — the 'one boat' memo promised 100% participation; does the retained team's output at devin desktop vindicate the six-day memo?"
    }

18. can the ioi-pipeline moat compound — does competitive-programming skill keep transferring to agent systems, or does the model layer commoditize the advantage? · cognition-idea-open-questions

    Original location

    Original structured fields

    {
      "text": "can the ioi-pipeline moat compound — does competitive-programming skill keep transferring to agent systems, or does the model layer commoditize the advantage?"
    }

19. what happens to the 'is devin real' question when the answer is a $48b company — does the discourse update, or does the demo-reality gap just get amortized? · cognition-idea-open-questions

    Original location

    Original structured fields

    {
      "text": "what happens to the 'is devin real' question when the answer is a $48b company — does the discourse update, or does the demo-reality gap just get amortized?"
    }

20. does outcome pricing hold — the productivity guarantee is the strongest version of 'agents are labor'; does the self-measured estimator survive a contested claim? · cognition-idea-open-questions

    Original location

    Original structured fields

    {
      "text": "does outcome pricing hold — the productivity guarantee is the strongest version of 'agents are labor'; does the self-measured estimator survive a contested claim?"
    }

21. who is customer no. 1 in the next category — does the government/genesis-mission work make devin critical infrastructure before the discourse notices? · cognition-idea-open-questions

    Original location

    Original structured fields

    {
      "text": "who is customer no. 1 in the next category — does the government/genesis-mission work make devin critical infrastructure before the discourse notices?"
    }

22. naming the category · cognition-idea-proved

    Original location · source-24b7f0d0ba5fb7d9ce77 · source-ab68c048f622c5487cd3 · source-64ac502a1459aa6bdd73 · source-60f035c084f29a27c33d · source-75343f5d5daaea74a9ea

    Original structured fields

    {
      "detail": "the category got named — 'the first ai software engineer' was a claim, not a description, but it made 'coding agent' a category overnight. by 2025–26 every major lab ships one under devin's framing: codex, jules, copilot coding agent, claude code. the naming is the durable win.",
      "id": "clm-category-named",
      "sourceIds": [
        "src-hp-cognition-introducing-devin",
        "src-ci-openai-codex",
        "src-ci-google-jules",
        "src-ci-copilot-launch",
        "src-ci-aicost-claudecode"
      ],
      "title": "naming the category"
    }

23. agent work sells · cognition-idea-proved

    Original location · source-27be554657a19b7483e8 · source-9c8b8da0c4acd715460c · source-6605de8983b4cf7a104f · source-f1e10ae59149bbf19428

    Original structured fields

    {
      "detail": "agents can do real billable-shaped work — not reliably enough to justify 'software engineer,' but enough to run a business: goldman's hundreds-of-devins deployment, the '16x' timeline claims, a $20 entry plan that works because the failure modes are priced in rather than eliminated.",
      "id": "clm-agent-work-sells",
      "sourceIds": [
        "src-ci-cnbc-goldman",
        "src-ci-fortune-goldman",
        "src-hp-cognition-ga",
        "src-ci-tc-devin2"
      ],
      "title": "agent work sells"
    }

24. the valuation arc · cognition-idea-proved

    Original location · source-e89e502cd59154c94c6f · source-edd94c292c5fa0292bc5 · source-2f9fa33e2b5ef408d29f · source-0760f59db9d9b7b278b7 · source-0db751b75021fab92d04

    Original structured fields

    {
      "detail": "a 2024 launch company can ride the agent wave to frontier-adjacent scale — $350m → $2b → $4b → $10.2b → $26b → $48b in ~30 months, with in-house models (swe-1.x, swe-2, swe-check) served by cerebras. the valuation arc is the strongest evidence the bet worked — all figures reported, most self-reported at the company layer.",
      "id": "clm-arc-worked",
      "sourceIds": [
        "src-hp-bloomberg-4b",
        "src-ci-tc-10b",
        "src-ci-tc-26b",
        "src-hp-reuters-48b",
        "src-ci-cognition-swe15"
      ],
      "title": "the valuation arc"
    }

25. the windsurf conversion · cognition-idea-proved

    Original location · source-b5c30eea0d72caf8fa23 · source-6de360c4af0fe6cbb866 · source-ce62ab0d51a90b48ed1f · source-b2840ab1745c1bf7e839

    Original structured fields

    {
      "detail": "distribution beats capability at the margin — the windsurf deal converted a dying acquisition target into an ide channel, a customer list, and $82m of arr in a weekend; by april 2026 'devin in windsurf' and june 2026 devin desktop made the acquisition the product, not the press release.",
      "id": "clm-distribution",
      "sourceIds": [
        "src-ci-cognition-windsurf",
        "src-ci-tc-cognition-windsurf",
        "src-ci-cognition-devin-windsurf",
        "src-ci-cognition-devin-desktop"
      ],
      "title": "the windsurf conversion"
    }

26. the labor frame held · cognition-idea-proved

    Original location · source-24b7f0d0ba5fb7d9ce77 · source-27be554657a19b7483e8 · source-8d6eed5ab8be721ab011 · source-e6d3a1b178c2d535cf22

    Original structured fields

    {
      "detail": "the labor framing survived the debunk — 'hire devin,' 'like our new employee,' 'a full operating ai software engineer,' engineer-hour guarantees: cognition kept selling labor-shaped software while discourse argued about whether it could. the guarantee even prices the claim.",
      "id": "clm-labor-frame",
      "sourceIds": [
        "src-hp-cognition-introducing-devin",
        "src-ci-cnbc-goldman",
        "src-hp-scottwu-linkedin-post",
        "src-ci-cognition-guarantee"
      ],
      "title": "the labor frame held"
    }

27. the gap stayed a gap · cognition-idea-stalled

    Original location · source-3e2dace97d27f013de2c · source-0db751b75021fab92d04 · source-f1e10ae59149bbf19428 · source-034c7082737b3e1133af

    Original structured fields

    {
      "detail": "the reliability gap never closed on the original terms — answer.ai's 3-of-20 stands as the independent benchmark; the 'unassisted 13.86%' never got a like-for-like reproduction; cognition's answer was to stop reporting swe-bench and to lower the price until 'sometimes wrong' was affordable, not to close the gap.",
      "id": "clm-reliability-gap",
      "sourceIds": [
        "src-ci-answerai-devin",
        "src-ci-cognition-swe15",
        "src-ci-tc-devin2",
        "src-ci-plainai-devin2"
      ],
      "title": "the gap stayed a gap"
    }

28. autonomy receded into supervision · cognition-idea-stalled

    Original location · source-4f32044a32d07efe7de4 · source-69d3c314b0e03d32799f · source-6605de8983b4cf7a104f

    Original structured fields

    {
      "detail": "the 'fully autonomous' claim stalled into supervision — devin 2.x added interactive planning, confidence indicators, devin review, and 'devins managing devins' — the product itself migrating from 'replace the engineer' toward 'augment the reviewer.' the launch's strongest claim became the roadmap's quietest admission.",
      "id": "clm-autonomy-receded",
      "sourceIds": [
        "src-ci-devin2",
        "src-hp-cognition-blog-index",
        "src-hp-cognition-ga"
      ],
      "title": "autonomy receded into supervision"
    }

29. no trusted ruler · cognition-idea-stalled

    Original location · source-7fb1ba94cb65b0e117ca · source-d0e3c86f1a860f403712 · source-e7e24fdf9169558ae8eb · source-3626bb632e52d9515da5

    Original structured fields

    {
      "detail": "the benchmark problem went unsolved — swe-bench's authority decayed (leakage, subset shopping, vendor self-evals) without a replacement ruler; by 2026 every agent company grades its own homework and the 'how good is it really' question the launch raised is still unanswered in public.",
      "id": "clm-benchmark-unsolved",
      "sourceIds": [
        "src-ci-swebench-plus",
        "src-ci-swebench-verified",
        "src-ci-agentmarketcap-swebench",
        "src-ci-digitalapplied-swebench"
      ],
      "title": "no trusted ruler"
    }

30. opaque agent economics · cognition-idea-stalled

    Original location · source-9d13db24b6d98238da40 · source-8aee4983e8bd1c160df7 · source-e6d3a1b178c2d535cf22 · source-5385b4f4e18becb33bbf

    Original structured fields

    {
      "detail": "the unit economics stay opaque — acus, core plans, run-rate revenue, 'net burn under $20m,' 3–4x productivity claims, the guarantee's estimator: the metrics are real business data but all company-published; nobody outside cognition can compute the margin on an engineer-hour.",
      "id": "clm-economics-opaque",
      "sourceIds": [
        "src-ci-productgeeks-pricing",
        "src-hp-cognition-series-c",
        "src-ci-cognition-guarantee",
        "src-ci-tc-48b"
      ],
      "title": "opaque agent economics"
    }

31. openai codex · cognition-idea-successors

    Original location · source-ab68c048f622c5487cd3 · source-02bdccaebbf3168398e0

    Original structured fields

    {
      "href": "https://openai.com/codex",
      "id": "suc-codex",
      "name": "openai codex",
      "note": "the foundation-model answer — chatgpt's coding agent, launched into the post-devin category in 2025; the platform-underneath threat cognition's 'independence' is priced against.",
      "sourceIds": [
        "src-ci-openai-codex",
        "src-ci-tc-codex"
      ]
    }

32. google jules + deepmind · cognition-idea-successors

    Original location · source-64ac502a1459aa6bdd73 · source-b87d95c58f8c4efe90a7

    Original structured fields

    {
      "href": "https://jules.google",
      "id": "suc-jules",
      "name": "google jules + deepmind",
      "note": "google's async coding agent — plus the windsurf founders inside deepmind working on agentic coding for gemini. the deal that made cognition also made its most dangerous competitor.",
      "sourceIds": [
        "src-ci-google-jules",
        "src-hp-reuters-google-windsurf"
      ]
    }

33. github copilot coding agent · cognition-idea-successors

    Original location · source-60f035c084f29a27c33d

    Original structured fields

    {
      "href": "https://github.com/features/copilot",
      "id": "suc-copilot-agent",
      "name": "github copilot coding agent",
      "note": "the incumbent's pivot — copilot grew an agent mode and an async coding agent; the 'copilot' brand now covers the thing devin defined against it.",
      "sourceIds": [
        "src-ci-copilot-launch"
      ]
    }

34. claude code · cognition-idea-successors

    Original location · source-75343f5d5daaea74a9ea · source-8e9720f44994fc564ffa · source-b5c30eea0d72caf8fa23

    Original structured fields

    {
      "href": "https://claude.com/product/claude-code",
      "id": "suc-claude-code",
      "name": "claude code",
      "note": "anthropic's agent — the terminal-native tool that repriced expectations, and the vendor whose claude access was the windsurf deal's hidden asset. cognition builds on it while competing with it.",
      "sourceIds": [
        "src-ci-aicost-claudecode",
        "src-ci-tc-anthropic-windsurf",
        "src-ci-cognition-windsurf"
      ]
    }

35. cursor · cognition-idea-successors

    Original location · source-5d96dd0653e8ac0fc447 · source-78cec34a9a8099f1b8ff · source-b7bf0d790557c9ef893c

    Original structured fields

    {
      "href": "https://cursor.com",
      "id": "suc-cursor",
      "name": "cursor",
      "note": "the ide-first rival — the mindshare leader in agentic editing, the customer-base overlap (spacex-scale logos on both sides), and walden yan's reported former employer. the fork in the road cognition's ide pivot was answering.",
      "sourceIds": [
        "src-ci-bnt-cursor-share",
        "src-ci-forbes-spacex-cursor",
        "src-hp-mungomash-leadership"
      ]
    }

36. openhands · cognition-idea-successors

    Original location · source-96df22d52de2b697c293

    Original structured fields

    {
      "href": "https://github.com/All-Hands-AI/OpenHands",
      "id": "suc-openhands",
      "name": "openhands",
      "note": "the open-source replication — created the day after the launch, ~85k stars; the standing 'we can just build it' answer and the control group for every capability claim.",
      "sourceIds": [
        "src-ci-opendevin-repo"
      ]
    }