Goal
Produce a deterministic shortlist URL that captures which use case, which verified-field filters, and which model order the catalogue surfaced.
Prerequisites
- Read /learn/how-to-choose-ai-model so the workflow framing makes sense.
- Know which workload you are evaluating (long-context, multimodal, hosted, governance, or another existing use case).
Step-by-step exercise
Pick a use case
Open the use cases hub and choose the one that matches the workload you are evaluating.
Expected outcome: You can name which verified fields the use case asks the catalogue to weight.
Open the selection workspace with your use case pre-selected
Visit /select and set the Use case filter, then narrow further by lifecycle (active only) and verified pricing coverage if relevant.
Expected outcome: The URL bar now carries your filter selections — copy this URL, it is your shortlist artifact.
Read the deterministic order
The shortlist orders by verified field count, active lifecycle, source count, name — never by score. Read the rows top to bottom.
Expected outcome: You can describe out loud why each row is ordered where it is, without invoking quality or price as the reason.
Open a top-three candidate model page
Click any model name in the shortlist. Read the verified-field rows and the citation count.
Expected outcome: You can name at least one data gap (an unverified field) for that model.
Completion checklist
Completion checklist
- The shortlist URL is saved.
- You can describe why the shortlist ordered itself the way it did.
- You named at least one data gap on the top candidate.
- You did NOT pick a winner.
Evidence artifact
At the end of this exercise you should have: A /select URL with your use-case + filter parameters that opens the same shortlist for any teammate.
Paste the artifact into your design doc, ticket, or PR description. The catalogue's role ends with the artifact; the workload-specific testing is yours.
Related workflow routes
- /select — narrow the source-backed shortlist.
- /compare/build — render verified fields side by side.
- /briefs/build — generate the evidence decision brief.
- /sources — every primary-source citation, by provider.
- /coverage — per-provider verified-field coverage.
- /reverification — sources due for manual re-check.
Exercise does not recommend a model — external testing still required.
Common mistake
Treating the first row of the shortlist as the answer. Shortlist order is deterministic (verified field count → lifecycle → source count → name), not a quality ranking.
Example artifact
Illustrative example — not a recommendation. Substitute your own values when you run the workflow.
Example artifact
## Shortlist URL (illustrative)
/select?useCase=<your-use-case>&lifecycle=active&pricingCoverage=verified
Top 3 rows (in catalogue order):
- <candidate-A> · verified context: <tokens> · active
- <candidate-B> · verified context: <tokens> · active
- <candidate-C> · verified context: <tokens> · active
Data gap noted: <candidate-C> has unverified max-output. Plan to confirm externally.Substitute your real shortlist, slugs, dates, and values when you walk the exercise.
Repeat this exercise when
- The use case or workload changes substantially.
- A snapshot rotates and lifecycle fields shift.
- Pricing references go stale and need re-verification.
- A teammate needs to inherit the shortlist URL.
Review before moving on
- Shortlist URL is captured (copy from the address bar).
- I can describe ordering rationale without using quality language.
- At least one data gap is named on the top candidate.
- No winner is declared in the notes.
Caution: No persistence — checklist resets on every visit. Capture progress in your own notes.
What this exercise does not produce
- A model recommendation. The exercise routes you through evidence — you decide.
- A score or grade for any model. The catalogue does not score.
- A substitute for external prompt, latency, rate-limit, cost, or compliance tests.