{
  "apiVersion": "v1",
  "topic": {
    "id": "choosing-models",
    "title": "How do organizations choose and switch AI models?",
    "group": "Tools, models and knowledge",
    "summary": "How organizations pick among frontier, open and fine-tuned models, avoid lock-in to one provider, and test before switching.",
    "passages": 15,
    "url": "https://the-ai-record.com/topics/choosing-models",
    "json": "https://the-ai-record.com/api/v1/topics/choosing-models.json"
  },
  "passages": [
    {
      "id": "airbnb-fine-tuning-gemma4-e2b-lifts-f1-2026",
      "text": "Fine-tuning Gemma4-E2B on hundreds of thousands of domain-specific examples lifts its F1 from 0.563 to 0.902, on par with the best proprietary models, while median latency stays at 35ms.",
      "context": null,
      "speaker": null,
      "speakerId": null,
      "speakerUrl": null,
      "speakerLevel": null,
      "role": null,
      "voice": "organization",
      "organization": "Airbnb",
      "organizationId": "airbnb",
      "organizationUrl": "https://the-ai-record.com/organizations/airbnb",
      "aiVendor": false,
      "work": {
        "title": "Frontier quality at 1% of the cost: a production benchmark of LLM classifiers",
        "type": "post",
        "date": "2026-11-03",
        "publisher": "Airbnb Tech blog",
        "locator": "Quality vs. latency",
        "url": "https://airbnb.tech/ai-ml/frontier-quality-at-1-of-the-cost-a-production-benchmark-of-llm-classifiers/"
      },
      "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": [
        "measuring-impact",
        "internal-tools",
        "choosing-models",
        "ai-evals"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [
        {
          "metric": "f1-after-fine-tuning",
          "value": 0.902,
          "unit": "score",
          "period": "2026 benchmark",
          "definition": "max F1 of fine-tuned Gemma4-E2B on a held-out set of 10,000 weighted examples, up from 0.563 before fine-tuning",
          "level": "n/a"
        }
      ],
      "url": "https://the-ai-record.com/passages/airbnb-fine-tuning-gemma4-e2b-lifts-f1-2026",
      "json": "https://the-ai-record.com/api/v1/passages/airbnb-fine-tuning-gemma4-e2b-lifts-f1-2026.json"
    },
    {
      "id": "airbnb-proprietary-models-only-win-low-traffic-2026",
      "text": "Proprietary models only win at low traffic: below roughly 15 to 30 RPM, Claude Sonnet 5.5 and GPT-6.1 Sol are cheaper than a dedicated GPU.",
      "context": null,
      "speaker": null,
      "speakerId": null,
      "speakerUrl": null,
      "speakerLevel": null,
      "role": null,
      "voice": "organization",
      "organization": "Airbnb",
      "organizationId": "airbnb",
      "organizationUrl": "https://the-ai-record.com/organizations/airbnb",
      "aiVendor": false,
      "work": {
        "title": "Frontier quality at 1% of the cost: a production benchmark of LLM classifiers",
        "type": "post",
        "date": "2026-11-03",
        "publisher": "Airbnb Tech blog",
        "locator": "Cost at scale",
        "url": "https://airbnb.tech/ai-ml/frontier-quality-at-1-of-the-cost-a-production-benchmark-of-llm-classifiers/"
      },
      "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": [
        "ai-cost-and-spend",
        "controlling-ai-spend",
        "choosing-models"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/airbnb-proprietary-models-only-win-low-traffic-2026",
      "json": "https://the-ai-record.com/api/v1/passages/airbnb-proprietary-models-only-win-low-traffic-2026.json"
    },
    {
      "id": "notion-auto-model-handles-about-75-traffic-2026",
      "text": "Notion's auto model does this really well. We have state-of-the-art models available always. But we also have an auto model there at the top that handles about 75% of our traffic, right? We have the ability to switch between models and our product, and we also offer it to our customers … they have access to these models without vendor lock-in.",
      "context": "Sachs is describing how Notion avoids depending on a single AI model provider.",
      "speaker": "Sarah Sachs",
      "speakerId": "sarah-sachs",
      "speakerUrl": "https://the-ai-record.com/people/sarah-sachs",
      "speakerLevel": "leader",
      "role": "Leads engineering teams for AI, Notion",
      "voice": "person",
      "organization": "Notion",
      "organizationId": "notion",
      "organizationUrl": "https://the-ai-record.com/organizations/notion",
      "aiVendor": true,
      "work": {
        "title": "Notion's Token Town",
        "type": "talk",
        "date": "2026-07-23",
        "publisher": "AI Engineer (YouTube)",
        "locator": "13:15",
        "url": "https://www.youtube.com/watch?v=-I5W5QVAT8E"
      },
      "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": "2026-11-04"
      },
      "topics": [
        "ai-cost-and-spend",
        "product-and-design",
        "controlling-ai-spend",
        "choosing-models"
      ],
      "functions": [
        "product",
        "engineering"
      ],
      "figures": [
        {
          "metric": "auto-model-traffic-share",
          "value": 75,
          "unit": "percent",
          "period": "stated in the 2026 talk",
          "definition": "Share of Notion's traffic that its auto model handles, as Sachs says it: 'about 75% of our traffic'.",
          "level": "unclear"
        }
      ],
      "url": "https://the-ai-record.com/passages/notion-auto-model-handles-about-75-traffic-2026",
      "json": "https://the-ai-record.com/api/v1/passages/notion-auto-model-handles-about-75-traffic-2026.json"
    },
    {
      "id": "shopify-now-number-distilled-models-where-take-teacher-2026",
      "text": "we now have a number of distilled models where we take a teacher model, usually a big frontier model, then teach a specific use case to a smaller model",
      "context": null,
      "speaker": "Harley Finkelstein",
      "speakerId": "harley-finkelstein",
      "speakerUrl": "https://the-ai-record.com/people/harley-finkelstein",
      "speakerLevel": "executive",
      "role": "President",
      "voice": "person",
      "organization": "Shopify",
      "organizationId": "shopify",
      "organizationUrl": "https://the-ai-record.com/organizations/shopify",
      "aiVendor": false,
      "work": {
        "title": "earnings call transcript via Platform Aeronaut (third-party transcript site)",
        "type": "earnings-call",
        "date": "2026-08-05",
        "publisher": "earnings call transcript via Platform Aeronaut (third-party transcript site)",
        "url": "https://transcripts.platformaeronaut.com/transcripts/SHOP-2Q26-transcript"
      },
      "grade": "A",
      "gradeMeaning": "A filing or earnings call the organization answers for.",
      "verification": {
        "status": "verified",
        "meaning": "An independent second reader re-fetched the source and confirmed the wording.",
        "reviewedOn": "2026-10-31"
      },
      "topics": [
        "ai-across-the-company",
        "choosing-models"
      ],
      "functions": [
        "company-wide"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/shopify-now-number-distilled-models-where-take-teacher-2026",
      "json": "https://the-ai-record.com/api/v1/passages/shopify-now-number-distilled-models-where-take-teacher-2026.json"
    },
    {
      "id": "airbnb-glowc-fine-tuned-classifier-matches-frontier-quality-2026",
      "text": "GLOWC, an internal framework used by engineers at Airbnb, makes it easy to fine-tune open-weight language (and vision-language) models into fast, accurate classifiers. In this blog post, we put one of these classifiers head-to-head with open-weight and proprietary models and find that it matches frontier-level quality in under 100ms at roughly 1% of the cost.",
      "context": null,
      "speaker": null,
      "speakerId": null,
      "speakerUrl": null,
      "speakerLevel": null,
      "role": null,
      "voice": "organization",
      "organization": "Airbnb",
      "organizationId": "airbnb",
      "organizationUrl": "https://the-ai-record.com/organizations/airbnb",
      "aiVendor": false,
      "work": {
        "title": "Frontier quality at 1% of the cost: a production benchmark of LLM classifiers",
        "type": "post",
        "date": "2026-11-03",
        "publisher": "Airbnb Tech blog",
        "locator": "Introduction",
        "url": "https://airbnb.tech/ai-ml/frontier-quality-at-1-of-the-cost-a-production-benchmark-of-llm-classifiers/"
      },
      "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": [
        "internal-tools",
        "ai-cost-and-spend",
        "controlling-ai-spend",
        "choosing-models"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [
        {
          "metric": "classifier-cost-vs-frontier",
          "value": 0.01,
          "unit": "share",
          "period": "2026 benchmark",
          "definition": "approximate cost of the fine-tuned classifier relative to frontier models on one production policy task, at public list rates and single-GPU rental cost",
          "level": "human-out"
        }
      ],
      "url": "https://the-ai-record.com/passages/airbnb-glowc-fine-tuned-classifier-matches-frontier-quality-2026",
      "json": "https://the-ai-record.com/api/v1/passages/airbnb-glowc-fine-tuned-classifier-matches-frontier-quality-2026.json"
    },
    {
      "id": "airbnb-labeled-data-examples-teach-distinctions-prompt-2026",
      "text": "You have labeled data. Examples teach distinctions and edge cases that instructions struggle to spell out. Hundreds of thousands of in-domain examples did what a 2,700 word prompt couldn’t.",
      "context": null,
      "speaker": null,
      "speakerId": null,
      "speakerUrl": null,
      "speakerLevel": null,
      "role": null,
      "voice": "organization",
      "organization": "Airbnb",
      "organizationId": "airbnb",
      "organizationUrl": "https://the-ai-record.com/organizations/airbnb",
      "aiVendor": false,
      "work": {
        "title": "Frontier quality at 1% of the cost: a production benchmark of LLM classifiers",
        "type": "post",
        "date": "2026-11-03",
        "publisher": "Airbnb Tech blog",
        "locator": "When should you fine-tune?",
        "url": "https://airbnb.tech/ai-ml/frontier-quality-at-1-of-the-cost-a-production-benchmark-of-llm-classifiers/"
      },
      "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": [
        "internal-tools",
        "choosing-models"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/airbnb-labeled-data-examples-teach-distinctions-prompt-2026",
      "json": "https://the-ai-record.com/api/v1/passages/airbnb-labeled-data-examples-teach-distinctions-prompt-2026.json"
    },
    {
      "id": "monzo-platform-provider-agnostic-by-definition-agent-box-swapped-any-model-2026",
      "text": "Our platform is provider agnostic by definition. So if we come back to the diagram, the little blue … green agent box there can be very easily swapped for any model. We can dynamically choose which model best fits the purposes that we have. So we are not specifically tied to any provider by definition",
      "context": "Felisola Caso is explaining why Monzo expects to keep its in-house agent platform as models change.",
      "speaker": "Felippe Felisola Caso",
      "speakerId": "felippe-felisola-caso",
      "speakerUrl": "https://the-ai-record.com/people/felippe-felisola-caso",
      "speakerLevel": "practitioner",
      "role": "Monzo (conference affiliation, 2026)",
      "voice": "person",
      "organization": "Monzo",
      "organizationId": "monzo",
      "organizationUrl": "https://the-ai-record.com/organizations/monzo",
      "aiVendor": false,
      "work": {
        "title": "A Bank Built Its Own Coding Agent. It Took Off (talk, AI Engineer)",
        "type": "talk",
        "date": "2026-10-20",
        "publisher": "AI Engineer (conference organizer) YouTube channel",
        "locator": "13:46",
        "url": "https://www.youtube.com/watch?v=adfw3xJBx1o"
      },
      "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": "2026-11-06"
      },
      "topics": [
        "internal-tools",
        "choosing-models"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/monzo-platform-provider-agnostic-by-definition-agent-box-swapped-any-model-2026",
      "json": "https://the-ai-record.com/api/v1/passages/monzo-platform-provider-agnostic-by-definition-agent-box-swapped-any-model-2026.json"
    },
    {
      "id": "cloudflare-later-randomized-split-opus-55-medium-2026",
      "text": "In a later randomized split, Opus 5.5 at medium showed no measurable change in review outcomes or critical findings compared with Opus 5, so we switched.",
      "context": null,
      "speaker": "Ryan Skidmore",
      "speakerId": "ryan-skidmore",
      "speakerUrl": "https://the-ai-record.com/people/ryan-skidmore",
      "speakerLevel": "practitioner",
      "role": "Author of Cloudflare's \"Orchestrating AI Code Review at scale\" blog post",
      "voice": "person",
      "organization": "Cloudflare",
      "organizationId": "cloudflare",
      "organizationUrl": "https://the-ai-record.com/organizations/cloudflare",
      "aiVendor": true,
      "work": {
        "title": "Cutting our inference bill by 68% by being obsessive about optimization",
        "type": "post",
        "date": "2026-10-19",
        "publisher": "Cloudflare blog",
        "locator": "Changing models and effort",
        "url": "https://blog.cloudflare.com/obsessive-inference-optimization/"
      },
      "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": [
        "choosing-models",
        "ai-evals"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/cloudflare-later-randomized-split-opus-55-medium-2026",
      "json": "https://the-ai-record.com/api/v1/passages/cloudflare-later-randomized-split-opus-55-medium-2026.json"
    },
    {
      "id": "uber-800-projects-100-million-model-requests-per-day-2026",
      "text": "Today, we have 800 plus projects internally going through this, cumulatively handling more than 100 million model requests per day. This includes both the frontier models and also open source models.",
      "context": "Medisetty is describing Uber's model gateway, the single endpoint all internal model use goes through.",
      "speaker": "Uday Kiran Medisetty",
      "speakerId": "uday-kiran-medisetty",
      "speakerUrl": "https://the-ai-record.com/people/uday-kiran-medisetty",
      "speakerLevel": "practitioner",
      "role": "Distinguished Engineer, Uber",
      "voice": "person",
      "organization": "Uber",
      "organizationId": "uber",
      "organizationUrl": "https://the-ai-record.com/organizations/uber",
      "aiVendor": false,
      "work": {
        "title": "Agentic SDLC at Uber: Building Blocks for Uber's Software Factory",
        "type": "talk",
        "date": "2026-08-21",
        "publisher": "AI Engineer (YouTube)",
        "locator": "03:30",
        "url": "https://www.youtube.com/watch?v=17-YSUHo6Lk"
      },
      "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": [
        "adoption-and-usage",
        "internal-tools",
        "choosing-models"
      ],
      "functions": [
        "engineering",
        "company-wide"
      ],
      "figures": [
        {
          "metric": "model-requests-per-day",
          "value": 100000000,
          "unit": "requests per day",
          "period": "at the time of the talk, 2026",
          "definition": "Model requests handled cumulatively by Uber's internal model gateway, frontier and open source models (more than)",
          "level": "n/a"
        },
        {
          "metric": "gateway-projects",
          "value": 800,
          "unit": "projects",
          "period": "at the time of the talk, 2026",
          "definition": "Internal projects going through Uber's model gateway (800 plus)",
          "level": "n/a"
        }
      ],
      "url": "https://the-ai-record.com/passages/uber-800-projects-100-million-model-requests-per-day-2026",
      "json": "https://the-ai-record.com/api/v1/passages/uber-800-projects-100-million-model-requests-per-day-2026.json"
    },
    {
      "id": "xebia-planning-agent-runs-your-most-powerful-2026",
      "text": "The planning agent is responsible for the actual thinking, so it should run on your most powerful model. I initially used Fable for this purpose when it was available, as it excels in UI/UX design. When Fable was no longer accessible, I switched to Opus. Since the planner writes the specifications, the implementation can be handled by a more cost-effective and faster model like Sonnet.",
      "context": null,
      "speaker": "Noah Blauensteiner",
      "speakerId": "noah-blauensteiner",
      "speakerUrl": "https://the-ai-record.com/people/noah-blauensteiner",
      "speakerLevel": "practitioner",
      "role": "Author, Xebia articles",
      "voice": "person",
      "organization": "Xebia",
      "organizationId": "xebia",
      "organizationUrl": "https://the-ai-record.com/organizations/xebia",
      "aiVendor": true,
      "work": {
        "title": "Shining Some Light in Dark Agentic Software Factories",
        "type": "post",
        "date": "2026-08-12",
        "publisher": "Xebia articles",
        "locator": "What is an Agentic Software Factory?",
        "url": "https://xebia.com/articles/shining-some-light-in-dark-agentic-software-factories/"
      },
      "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": "2026-10-31"
      },
      "topics": [
        "coding-agents",
        "ai-cost-and-spend",
        "choosing-models"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/xebia-planning-agent-runs-your-most-powerful-2026",
      "json": "https://the-ai-record.com/api/v1/passages/xebia-planning-agent-runs-your-most-powerful-2026.json"
    },
    {
      "id": "notion-not-all-traffic-equal-triage-email-opus-2026",
      "text": "And not all traffic is equal. It is a huge miss to send all of these to the latest Opus model. Some of these absolutely. Large scale data analysis, when you do it on Notion, will recommend Opus, right? When you triage an email inbox, if we're charging you to do that on Opus, we're ripping you off, and ourselves.",
      "context": "Sachs is arguing that an AI product should route different kinds of requests to different models.",
      "speaker": "Sarah Sachs",
      "speakerId": "sarah-sachs",
      "speakerUrl": "https://the-ai-record.com/people/sarah-sachs",
      "speakerLevel": "leader",
      "role": "Leads engineering teams for AI, Notion",
      "voice": "person",
      "organization": "Notion",
      "organizationId": "notion",
      "organizationUrl": "https://the-ai-record.com/organizations/notion",
      "aiVendor": true,
      "work": {
        "title": "Notion's Token Town",
        "type": "talk",
        "date": "2026-07-23",
        "publisher": "AI Engineer (YouTube)",
        "locator": "10:24",
        "url": "https://www.youtube.com/watch?v=-I5W5QVAT8E"
      },
      "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": "2026-11-04"
      },
      "topics": [
        "ai-cost-and-spend",
        "product-and-design",
        "controlling-ai-spend",
        "choosing-models"
      ],
      "functions": [
        "product",
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/notion-not-all-traffic-equal-triage-email-opus-2026",
      "json": "https://the-ai-record.com/api/v1/passages/notion-not-all-traffic-equal-triage-email-opus-2026.json"
    },
    {
      "id": "uber-every-model-request-goes-through-one-gateway-2026",
      "text": "And since all the requests go through one place, we could also add like audit logging, session traces, and a whole bunch of infrastructure around this. Every model request at Uber goes through this. So there's no model request that doesn't go through this. So that gives us like one single entry point for across all the models.",
      "context": "'This' is Uber's internal model gateway, which he has just been describing (one OpenAI-compatible or Anthropic-compatible API in front of models from all vendors, with an AI Guard safety layer).",
      "speaker": "Uday Kiran Medisetty",
      "speakerId": "uday-kiran-medisetty",
      "speakerUrl": "https://the-ai-record.com/people/uday-kiran-medisetty",
      "speakerLevel": "practitioner",
      "role": "Distinguished Engineer, Uber",
      "voice": "person",
      "organization": "Uber",
      "organizationId": "uber",
      "organizationUrl": "https://the-ai-record.com/organizations/uber",
      "aiVendor": false,
      "work": {
        "title": "Uber's AI infra, from the inside",
        "type": "podcast",
        "date": "2026-10-14",
        "publisher": "The Pragmatic Engineer (YouTube)",
        "locator": "24:22",
        "url": "https://www.youtube.com/watch?v=MDUYSMJmi8A"
      },
      "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": "2026-11-07"
      },
      "topics": [
        "internal-tools",
        "choosing-models",
        "controlling-ai-spend"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/uber-every-model-request-goes-through-one-gateway-2026",
      "json": "https://the-ai-record.com/api/v1/passages/uber-every-model-request-goes-through-one-gateway-2026.json"
    },
    {
      "id": "dropbox-build-agent-loop-tools-context-ourselves-framework-lagged-2026",
      "text": "Ultimately, we opted to build the agent loop, tools, and the system for gathering context ourselves, and we connect directly to model providers instead of relying on a broad agent framework. That's because a framework we initially explored lagged behind the provider APIs we wanted to use and included more structure than Reclaim needed.",
      "context": null,
      "speaker": null,
      "speakerId": null,
      "speakerUrl": null,
      "speakerLevel": null,
      "role": null,
      "voice": "organization",
      "organization": "Dropbox",
      "organizationId": "dropbox",
      "organizationUrl": "https://the-ai-record.com/organizations/dropbox",
      "aiVendor": false,
      "work": {
        "title": "Evolving our calendar assistant Reclaim to be AI-native without starting over",
        "type": "post",
        "date": "2026-09-29",
        "publisher": "Dropbox.Tech",
        "locator": "Building the agent platform",
        "url": "https://dropbox.tech/machine-learning/evolving-calendar-assistant-reclaim-to-be-ai-native"
      },
      "grade": "B",
      "gradeMeaning": "The organization's own post, paper or talk.",
      "verification": {
        "status": "reported",
        "meaning": "Not yet confirmed against the source by an independent second reader."
      },
      "topics": [
        "context-for-agents",
        "choosing-models"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/dropbox-build-agent-loop-tools-context-ourselves-framework-lagged-2026",
      "json": "https://the-ai-record.com/api/v1/passages/dropbox-build-agent-loop-tools-context-ourselves-framework-lagged-2026.json"
    },
    {
      "id": "cursor-bench-private-eval-own-codebase-2026",
      "text": "And that's why we have Cursor Bench. We have this private eval set that is mostly made up of things that happen in our code base, which is held out from the … So we ensure that the models aren't trained on it. And it's based on those real world engineering tasks.",
      "context": "Robinson has just said public benchmarks are an imperfect test of how models are used in practice, and is describing Cursor's private evaluation set for its own models.",
      "speaker": "Lee Robinson",
      "speakerId": "lee-robinson",
      "speakerUrl": "https://the-ai-record.com/people/lee-robinson",
      "speakerLevel": "practitioner",
      "role": "Machine Learning Engineer, Model Behavior, Cursor",
      "voice": "person",
      "organization": "Cursor (Anysphere)",
      "organizationId": "cursor",
      "organizationUrl": "https://the-ai-record.com/organizations/cursor",
      "aiVendor": true,
      "work": {
        "title": "Recursive Model Improvement",
        "type": "talk",
        "date": "2026-07-15",
        "publisher": "AI Engineer (YouTube)",
        "locator": "8:12",
        "url": "https://www.youtube.com/watch?v=q4Tr-DknG2M"
      },
      "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": [
        "measuring-impact",
        "choosing-models",
        "ai-evals"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/cursor-bench-private-eval-own-codebase-2026",
      "json": "https://the-ai-record.com/api/v1/passages/cursor-bench-private-eval-own-codebase-2026.json"
    },
    {
      "id": "nordic-corporate-bank-three-model-review-pool-2026",
      "text": "And then the self review, which is kind of interesting because we have this, we created this review pool. So we have one model, which is a Sonnet that does one review. … then we have kind of a … cheaper model that also does a review. And then we have the third model, … right now, it kind of consolidates this and verifies the findings in the reviews and decides if it should change something or not. And when this review pool is kind of finished and happy, … will assign it to a human.",
      "context": "The speaker describes how the agent reviews its own pull request with three models before handing it to a person.",
      "speaker": "Hallstein Brøtan",
      "speakerId": "hallstein-brotan",
      "speakerUrl": "https://the-ai-record.com/people/hallstein-brotan",
      "speakerLevel": "practitioner",
      "role": "Senior consultant, Novanet; developer on Nordic Corporate Bank's system",
      "voice": "person",
      "organization": "Nordic Corporate Bank",
      "organizationId": "nordic-corporate-bank",
      "organizationUrl": "https://the-ai-record.com/organizations/nordic-corporate-bank",
      "aiVendor": false,
      "work": {
        "title": "How We Built an Autonomous Development Workflow at Nordic Corporate Bank",
        "type": "talk",
        "date": "2026-06-29",
        "publisher": "NDC Conferences (YouTube); talk recorded at NDC in Oslo",
        "locator": "37:00",
        "url": "https://www.youtube.com/watch?v=p8M8--12h1Y"
      },
      "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": [
        "code-review",
        "ai-review-quality",
        "choosing-models"
      ],
      "functions": [
        "engineering"
      ],
      "figures": [],
      "url": "https://the-ai-record.com/passages/nordic-corporate-bank-three-model-review-pool-2026",
      "json": "https://the-ai-record.com/api/v1/passages/nordic-corporate-bank-three-model-review-pool-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."
  ]
}