Skip to content
WebmasterID

Start here · Engineers preparing an integration

Start here — Developer

Preparing an integration. Walk the developer path, then open the developer model evaluation kit to bundle lessons, exercises, lab playbook, prompt sets, and templates into a single Markdown work document.

Role

Developer

First step

Walk the developer learning path.

Estimated time

~180 min

3-minute orientation

Skim this before opening the first lesson. Estimated time for the full role path: ~180 minutes.

Learn
Read how-to-choose-ai-model, hosted-vs-first-party, structured-output, testing-ai-models.
Practise
Build the comparison and capture a hosted-provider note via the exercises.
Test
Run prompt-testing-basics + structured-output-testing in your own harness.
Produce
Export a Markdown decision brief paired with an external test plan.

First route list

Your first route list

Open these in order. Every route already exists in the product — Start Here is an entry point, not a parallel surface.

  1. Learning path

    Developer learning path

    Sequenced developer lessons and exercises.

    Open /learn/path/developer →

  2. First lesson

    How to choose an AI model

    Frame the workflow before reading per-field lessons.

    Open /learn/how-to-choose-ai-model →

  3. First exercise

    Map the hosted provider

    Separate creator + host + data gap for the candidate set.

    Open /learn/exercises/map-hosted-provider →

  4. Lab playbook

    Prompt testing basics

    Minimum repeatable prompt-test routine before integration.

    Open /lab/prompt-testing-basics →

  5. Workflow kit

    Developer model evaluation kit

    Single Markdown work document bundling the developer flow.

    Open /kits/developer-model-evaluation →

  6. Resource finder

    Open developer resources

    Every resource tagged for developers.

    Open /resources?audience=developers →

What you will produce

Completion is the artifacts the path puts in your hands — not a certificate, badge, or progress bar.

What this does not promise

  • Pick the right model for the integration.
  • Assert latency, throughput, or uptime.
  • Replace workload-specific testing.
  • Certify the model for any regulatory regime.

How Start Here works

  • No accounts, no sign-in, no quiz scoring. Picking a role opens links — it does not save state about the reader.
  • No progress tracking. There is no completion bar, no certificate, no badge. Completion is the artifacts the path puts in your hands.
  • No personalization. Two readers who click the same card see the same content; no recommendation engine runs against the reader.
  • No model recommendations. The role pages route into learning workflows; which model fits the workload is the reader's decision after the testing step.
  • No SEO ranking guarantees, no production-readiness guarantees, no compliance certification. Start Here is a navigation aid, not a promise about outcomes.

Next step

Open the first route

Start with the learning path, or jump to the kit / finder if you already know the shape of the work.