{
  "apiVersion": "v1",
  "topic": {
    "id": "failures-and-reversals",
    "title": "What AI efforts failed or were rolled back?",
    "group": "Results and cost",
    "summary": "Tools turned off, experiments abandoned, policies reversed and results that came out negative.",
    "passages": 10,
    "url": "https://the-ai-record.com/topics/failures-and-reversals",
    "json": "https://the-ai-record.com/api/v1/topics/failures-and-reversals.json"
  },
  "passages": [
    {
      "id": "traderspost-built-ai-code-review-action-single-day-2026",
      "text": "We built an automated AI code review GitHub Action in a single day, and we reverted it before the day was over. The idea was simple: someone comments a trigger phrase on a pull request, and a workflow runs our deepest review and posts the result. It took seven small pull requests to get that far, and one revert pull request to undo them.",
      "context": null,
      "speaker": null,
      "speakerId": null,
      "speakerUrl": null,
      "speakerLevel": null,
      "role": null,
      "voice": "organization",
      "organization": "TradersPost",
      "organizationId": "traderspost",
      "organizationUrl": "https://the-ai-record.com/organizations/traderspost",
      "aiVendor": false,
      "work": {
        "title": "The AI Review Bot We Reverted",
        "type": "post",
        "date": "2026-10-13",
        "publisher": "TradersPost blog",
        "locator": "Introduction",
        "url": "https://blog.traderspost.io/article/the-ai-review-bot-we-built-and-reverted"
      },
      "grade": "B",
      "gradeMeaning": "The organization's own post, paper or talk.",
      "verification": {
        "status": "verified",
        "meaning": "An independent second reader re-fetched the source and confirmed the wording.",
        "reviewedOn": "2026-11-05"
      },
      "topics": [
        "code-review",
        "failures-and-reversals"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/traderspost-built-ai-code-review-action-single-day-2026",
      "json": "https://the-ai-record.com/api/v1/passages/traderspost-built-ai-code-review-action-single-day-2026.json"
    },
    {
      "id": "traderspost-nobody-wrote-postmortem-cannot-tell-noise-2026",
      "text": "Nobody wrote a postmortem. We did not collect a table of review comments the bot produced, measure their precision, or compare cost against the existing session-based review. So we cannot tell you the bot failed because of noise, or cost, or authentication friction, or a change of mind about where review should live.",
      "context": null,
      "speaker": null,
      "speakerId": null,
      "speakerUrl": null,
      "speakerLevel": null,
      "role": null,
      "voice": "organization",
      "organization": "TradersPost",
      "organizationId": "traderspost",
      "organizationUrl": "https://the-ai-record.com/organizations/traderspost",
      "aiVendor": false,
      "work": {
        "title": "The AI Review Bot We Reverted",
        "type": "post",
        "date": "2026-10-13",
        "publisher": "TradersPost blog",
        "locator": "What we do not know",
        "url": "https://blog.traderspost.io/article/the-ai-review-bot-we-built-and-reverted"
      },
      "grade": "B",
      "gradeMeaning": "The organization's own post, paper or talk.",
      "verification": {
        "status": "verified",
        "meaning": "An independent second reader re-fetched the source and confirmed the wording.",
        "reviewedOn": "2026-11-07"
      },
      "topics": [
        "code-review",
        "failures-and-reversals"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/traderspost-nobody-wrote-postmortem-cannot-tell-noise-2026",
      "json": "https://the-ai-record.com/api/v1/passages/traderspost-nobody-wrote-postmortem-cannot-tell-noise-2026.json"
    },
    {
      "id": "stripe-first-approach-orchestrator-agent-expensive-slow-2026",
      "text": "Our first approach used a high-level orchestrator agent for this role. It worked, but it was expensive, slow, and sometimes unreliable. … We learned that although an LLM can orchestrate a workflow, predictable coordination is better implemented in code.",
      "context": null,
      "speaker": null,
      "speakerId": null,
      "speakerUrl": null,
      "speakerLevel": null,
      "role": null,
      "voice": "organization",
      "organization": "Stripe",
      "organizationId": "stripe",
      "organizationUrl": "https://the-ai-record.com/organizations/stripe",
      "aiVendor": false,
      "work": {
        "title": "Stripe's Payment Method Factory: Orchestrating agents for repeated, custom integrations",
        "type": "post",
        "date": "2026-09-30",
        "publisher": "Stripe Dot Dev Blog (stripe.dev); authors David Dunne, Xenofon Vourliotis and Sai Samant",
        "locator": "Building the factory",
        "url": "https://stripe.dev/blog/stripes-payment-method-factory-orchestrating-agents-for-repeated-custom-integrations"
      },
      "grade": "B",
      "gradeMeaning": "The organization's own post, paper or talk.",
      "verification": {
        "status": "verified",
        "meaning": "An independent second reader re-fetched the source and confirmed the wording.",
        "reviewedOn": "2026-11-08"
      },
      "topics": [
        "failures-and-reversals",
        "parallel-agents"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/stripe-first-approach-orchestrator-agent-expensive-slow-2026",
      "json": "https://the-ai-record.com/api/v1/passages/stripe-first-approach-orchestrator-agent-expensive-slow-2026.json"
    },
    {
      "id": "ionix-agent-silently-lost-tools-answers-anyway-2026",
      "text": "An agent that has silently lost its tools does not report that it lost its tools. It answers anyway, with the same fluency as before.",
      "context": "From the post's account of a sandbox failure in which the review agent could not run any commands.",
      "speaker": "Arthur Vinberg",
      "speakerId": "arthur-vinberg",
      "speakerUrl": "https://the-ai-record.com/people/arthur-vinberg",
      "speakerLevel": "leader",
      "role": "Head of DevOps & IT",
      "voice": "person",
      "organization": "Ionix",
      "organizationId": "ionix",
      "organizationUrl": "https://the-ai-record.com/organizations/ionix",
      "aiVendor": false,
      "work": {
        "title": "We built an AI PR reviewer. The hard part was teaching it to say nothing.",
        "type": "post",
        "date": "2026-08-13",
        "publisher": "Ionix blog",
        "locator": "Two failures that taught us more than the design did",
        "url": "https://www.ionix.io/blog/we-built-an-ai-pr-reviewer/"
      },
      "grade": "B",
      "gradeMeaning": "The organization's own post, paper or talk.",
      "verification": {
        "status": "verified",
        "meaning": "An independent second reader re-fetched the source and confirmed the wording.",
        "reviewedOn": "2026-11-08"
      },
      "topics": [
        "failures-and-reversals",
        "autonomy-and-oversight"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/ionix-agent-silently-lost-tools-answers-anyway-2026",
      "json": "https://the-ai-record.com/api/v1/passages/ionix-agent-silently-lost-tools-answers-anyway-2026.json"
    },
    {
      "id": "boundary-ai-psychosis-ten-design-docs-day-read-rule-2026",
      "text": "But the real problem with all of this is I built this and I hit a little bit of AI psychosis. And I started shipping 10 design docs a day. And soon, the team was fighting my slop. … This last rule was … ship a design doc, you require people to actually go read it. And with this last standard, we suddenly had design docs … incredibly high quality.",
      "context": "Gupta is describing the design-document tool he built for the team and the rule that followed.",
      "speaker": "Vaibhav Gupta",
      "speakerId": "vaibhav-gupta",
      "speakerUrl": "https://the-ai-record.com/people/vaibhav-gupta",
      "speakerLevel": "executive",
      "role": "Co-founder and Chief Executive",
      "voice": "person",
      "organization": "Boundary (BoundaryML)",
      "organizationId": "boundary",
      "organizationUrl": "https://the-ai-record.com/organizations/boundary",
      "aiVendor": true,
      "work": {
        "title": "Fighting Slop with Slop",
        "type": "talk",
        "date": "2026-07-31",
        "publisher": "AI Engineer (YouTube)",
        "locator": "3:37",
        "url": "https://www.youtube.com/watch?v=AMiyLItEtLA"
      },
      "grade": "B",
      "gradeMeaning": "The organization's own post, paper or talk.",
      "verification": {
        "status": "verified",
        "meaning": "An independent second reader re-fetched the source and confirmed the wording.",
        "reviewedOn": "2026-11-01"
      },
      "topics": [
        "autonomy-and-oversight",
        "product-and-design",
        "failures-and-reversals"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/boundary-ai-psychosis-ten-design-docs-day-read-rule-2026",
      "json": "https://the-ai-record.com/api/v1/passages/boundary-ai-psychosis-ten-design-docs-day-read-rule-2026.json"
    },
    {
      "id": "humanlayer-daily-loop-pr-turned-off-high-friction-2026",
      "text": "And every morning, we walk into the office to a small incremental PR that's low risk. And when we first did this, it was actually really frustrating, and we turned the loop off … because we had to constantly update the skill, we had to constantly check out the branch, change the skill, change the code, commit and push, and our loop was actually really high friction, right?",
      "context": "Mistele has just described scheduling the migration loop to run once a day as a CI workflow, and goes on to the problem HumanLayer hit with the first version.",
      "speaker": "Kyle Mistele",
      "speakerId": "kyle-mistele",
      "speakerUrl": "https://the-ai-record.com/people/kyle-mistele",
      "speakerLevel": "executive",
      "role": "Co-founder, HumanLayer",
      "voice": "person",
      "organization": "HumanLayer",
      "organizationId": "humanlayer",
      "organizationUrl": "https://the-ai-record.com/organizations/humanlayer",
      "aiVendor": true,
      "work": {
        "title": "Loop Engineering from first principles",
        "type": "talk",
        "date": "2026-07-25",
        "publisher": "AI Engineer (YouTube)",
        "locator": "13:48",
        "url": "https://www.youtube.com/watch?v=xIt_mTQp6mY"
      },
      "grade": "B",
      "gradeMeaning": "The organization's own post, paper or talk.",
      "verification": {
        "status": "verified",
        "meaning": "An independent second reader re-fetched the source and confirmed the wording.",
        "reviewedOn": "2026-11-03"
      },
      "topics": [
        "coding-agents",
        "autonomy-and-oversight",
        "failures-and-reversals"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/humanlayer-daily-loop-pr-turned-off-high-friction-2026",
      "json": "https://the-ai-record.com/api/v1/passages/humanlayer-daily-loop-pr-turned-off-high-friction-2026.json"
    },
    {
      "id": "cursor-twenty-agents-slow-throughput-one-three-locks-2026",
      "text": "Locking also caused too much contention. 20 agents would slow to the throughput of 1-3 with most time spent waiting on locks.",
      "context": null,
      "speaker": "Wilson Lin",
      "speakerId": "wilson-lin",
      "speakerUrl": "https://the-ai-record.com/people/wilson-lin",
      "speakerLevel": "practitioner",
      "role": "Researcher, Cursor (author of the Towards self-driving codebases post)",
      "voice": "person",
      "organization": "Cursor (Anysphere)",
      "organizationId": "cursor",
      "organizationUrl": "https://the-ai-record.com/organizations/cursor",
      "aiVendor": true,
      "work": {
        "title": "Towards self-driving codebases",
        "type": "post",
        "date": "2026-02-05",
        "publisher": "Cursor (own site)",
        "locator": "Self-coordination",
        "url": "https://cursor.com/blog/self-driving-codebases"
      },
      "grade": "B",
      "gradeMeaning": "The organization's own post, paper or talk.",
      "verification": {
        "status": "verified",
        "meaning": "An independent second reader re-fetched the source and confirmed the wording.",
        "reviewedOn": "2026-11-08"
      },
      "topics": [
        "parallel-agents",
        "failures-and-reversals"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [
        {
          "metric": "agents-running",
          "value": 20,
          "unit": "agents",
          "period": "early multi-agent harness with a shared lock file",
          "definition": "agents sharing a coordination file with locks",
          "level": "human-out"
        }
      ],
      "url": "https://the-ai-record.com/passages/cursor-twenty-agents-slow-throughput-one-three-locks-2026",
      "json": "https://the-ai-record.com/api/v1/passages/cursor-twenty-agents-slow-throughput-one-three-locks-2026.json"
    },
    {
      "id": "cognition-vibe-coding-no-review-experiment-two-weeks-2026",
      "text": "I think we found that the state-of-the-art in December was you can probably, run this for about two weeks. By the end of those two weeks, you’d find that, hey, you want to, change the color of a button. Well, it turns out this button is implemented in, 10 different places, and they, have All these different variations, and oh, you forgot one of them, and actually it’s a slightly different color in one spot.",
      "context": "Yan was describing an internal experiment building real products by auto-merging AI-written code with no code review, to see how many weeks it could run before the codebase became unmanageable.",
      "speaker": "Walden Yan",
      "speakerId": "walden-yan",
      "speakerUrl": "https://the-ai-record.com/people/walden-yan",
      "speakerLevel": "executive",
      "role": "Co-founder and Chief Product Officer",
      "voice": "person",
      "organization": "Cognition",
      "organizationId": "cognition",
      "organizationUrl": "https://the-ai-record.com/organizations/cognition",
      "aiVendor": true,
      "work": {
        "title": "The Age of Async Agents: Cognition's Walden Yan and OpenInspect's Cole Murray",
        "type": "podcast",
        "date": "2026-05-28",
        "publisher": "Latent Space",
        "locator": "00:44:21",
        "url": "https://latent.space/p/cognition"
      },
      "grade": "C",
      "gradeMeaning": "An interview or press quote.",
      "verification": {
        "status": "verified",
        "meaning": "An independent second reader re-fetched the source and confirmed the wording.",
        "reviewedOn": "2026-11-01"
      },
      "topics": [
        "code-review",
        "autonomy-and-oversight",
        "reading-agent-code",
        "failures-and-reversals"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/cognition-vibe-coding-no-review-experiment-two-weeks-2026",
      "json": "https://the-ai-record.com/api/v1/passages/cognition-vibe-coding-no-review-experiment-two-weeks-2026.json"
    },
    {
      "id": "klarna-need-rethink-make-customer-service-human-experience-2026",
      "text": "We need to rethink this and make customer service a human experience. Everyone who wants a human connection should get it.",
      "context": null,
      "speaker": "Sebastian Siemiatkowski",
      "speakerId": "sebastian-siemiatkowski",
      "speakerUrl": "https://the-ai-record.com/people/sebastian-siemiatkowski",
      "speakerLevel": "executive",
      "role": "Chief Executive Officer and co-founder",
      "voice": "person",
      "organization": "Klarna",
      "organizationId": "klarna",
      "organizationUrl": "https://the-ai-record.com/organizations/klarna",
      "aiVendor": false,
      "work": {
        "title": "CX Foundation (specialist customer-experience media and analyst site)",
        "type": "interview",
        "date": "2026-02-17",
        "publisher": "CX Foundation (specialist customer-experience media and analyst site)",
        "url": "https://cxfoundation.com/news/klarna-recruits-customers-as-service-reps"
      },
      "grade": "C",
      "gradeMeaning": "An interview or press quote.",
      "verification": {
        "status": "verified",
        "meaning": "An independent second reader re-fetched the source and confirmed the wording.",
        "reviewedOn": "2026-10-31"
      },
      "topics": [
        "ai-support-resolution",
        "failures-and-reversals",
        "support-satisfaction"
      ],
      "functions": [
        "support"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/klarna-need-rethink-make-customer-service-human-experience-2026",
      "json": "https://the-ai-record.com/api/v1/passages/klarna-need-rethink-make-customer-service-human-experience-2026.json"
    },
    {
      "id": "traderspost-time-went-three-plumbing-problems-2026",
      "text": "The time went into three plumbing problems: what the action accepts as inputs, which permissions the job needs, and how the job authenticates to the model provider.",
      "context": null,
      "speaker": null,
      "speakerId": null,
      "speakerUrl": null,
      "speakerLevel": null,
      "role": null,
      "voice": "organization",
      "organization": "TradersPost",
      "organizationId": "traderspost",
      "organizationUrl": "https://the-ai-record.com/organizations/traderspost",
      "aiVendor": false,
      "work": {
        "title": "The AI Review Bot We Reverted",
        "type": "post",
        "date": "2026-10-13",
        "publisher": "TradersPost blog",
        "locator": "Seven pull requests in one day",
        "url": "https://blog.traderspost.io/article/the-ai-review-bot-we-built-and-reverted"
      },
      "grade": "B",
      "gradeMeaning": "The organization's own post, paper or talk.",
      "verification": {
        "status": "verified",
        "meaning": "An independent second reader re-fetched the source and confirmed the wording.",
        "reviewedOn": "2026-11-07"
      },
      "topics": [
        "code-review",
        "failures-and-reversals"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/traderspost-time-went-three-plumbing-problems-2026",
      "json": "https://the-ai-record.com/api/v1/passages/traderspost-time-went-three-plumbing-problems-2026.json"
    }
  ],
  "notice": [
    "Every passage is reproduced word for word from the cited source.",
    "Check the verification status: \"reported\" means not yet confirmed by an independent second reader.",
    "Organizations flagged aiVendor sell AI tools or services and have a reason to publish flattering numbers.",
    "An empty result means the archive holds nothing on it, not that nobody has said anything."
  ]
}