What AiModels WebmasterID is
AiModels WebmasterID is an AI usage learning platform powered by verified model intelligence. It packages four things into one product:
- A curriculum of plain-language lessons, practical exercises, and role-based learning paths.
- A workflow tier — selection workspace, comparison builder, decision brief builder — that produces paste-ready artifacts from verified catalogue rows.
- An AI Usage Lab with playbooks, Markdown templates, and evaluation prompt sets for testing model behaviour before integration.
- A verified-data backbone of models, providers, hosted availability, pricing references, source citations, status observations, and a reverification queue.
The product loop is short: Learn → Apply → Verify → Test. Every audience uses the same loop; the audience pages at /for only change the starting order.
What it is not
The platform is deliberately not several adjacent things that the AI ecosystem tends to confuse with it:
- Not an AI news feed. No daily roundups, no editorial commentary on launches. The catalogue records verified fields with timestamps instead.
- Not a prompt pack or marketplace. The evaluation prompt library is explicitly framed as evaluation inputs, never production prompts, and it is never ranked or sold.
- Not a model leaderboard or scoring engine. No model is declared better than another. Comparisons render verified fields side by side and leave the decision to the reader.
- Not a live pricing-quote engine. Pricing rows are sourced references with retrievedAt dates. They are not live invoiceable quotes.
- Not a certification authority. Verification means a primary source backed a value on the date recorded. It does not certify the model for any regulatory regime.
- Not an AI tools directory or affiliate site. No tool roundups, no affiliate links, no vendor endorsement.
Learn → Apply → Verify → Test
The product's core loop has four stages. Each stage has one canonical entry point:
- Learn — /learn. Concept lessons on context window, pricing references, hosted vs first-party, lifecycle, status, structured output, benchmark limits, modality, testing.
- Apply — /select, /compare/build, /briefs/build. Each workspace produces a deterministic URL or a Markdown artifact.
- Verify — /sources, /coverage, /reverification. Every claim links back to a primary source with a retrievedAt date.
- Test — /lab. Six playbooks, three Markdown templates, six evaluation prompt sets. Markdown exports always serve with
X-Robots-Tag: noindex.
Audience paths
Four audience entry points live under /for. They surface the same curriculum + workspaces in a different order:
- /for/developers — engineers preparing an integration.
- /for/product-teams — product managers / technical buyers turning a use case into a defensible decision.
- /for/automation-specialists — automation builders, SEO operators, technical consultants using AI models inside workflows.
- /for/governance-teams — risk, compliance, governance reviewers preparing internal approval discussions.
Evidence artifacts
Every walk through the platform ends with a concrete, paste-ready artifact. The catalogue never generates a score or a verdict; it produces evidence the reader's team owns the decision on.
- Shortlist URLs from /select.
- Comparison URLs from /compare/build.
- Markdown evidence briefs from /briefs/build (and the JSON variant for machine consumption).
- Model evaluation plans, prompt test matrices, and automation risk checklists from /lab/templates.
- Evaluation prompt sets from /lab/prompts for external test harness use.
- Source freshness checklists from /reverification and
/api/reverification/checklist.
Verified model intelligence backbone
The catalogue's verified-data layer is the load-bearing substrate for the rest of the product. Every metric (pricing, context window, max output, modality, knowledge cutoff, lifecycle, hosted availability) is either backed by a primary-source citation with a retrievedAt date, or it is omitted. Unverified fields render the canonical unverified-data label, never an estimate.
Primary sources are allow-listed: official vendor documentation, official pricing pages, regulatory filings, peer-reviewed papers, and public datasets. Blog posts, social media, and secondary summaries are not primary sources.
Lifecycle, freshness, and reverification are first-class concerns. The catalogue never auto-mutates verified values; the reverification queue surfaces sources due for manual re-check.
No rankings, no recommendations, no guarantees
Across every surface, the platform refuses to assert what the reader's team must decide:
- No "best model" or "winner" claims. Comparisons render verified fields without scores.
- No price ranking. Pricing rows are references with retrievedAt dates.
- No fabricated latency, throughput, or uptime. Status observations are recorded; SLA assertions are not.
- No benchmark publishing. Definitions are tracked; per-model scores require independently reproducible methodology before they land.
- No certification, no compliance approval, no production readiness guarantee, no SEO ranking guarantee. The lab policy says so explicitly.
- No accounts, no progress tracking, no badges, no course completion certificates.
How to use this platform responsibly
- Pick the audience page that matches your role at /for.
- Walk the recommended learning path to ground yourself in the relevant verified fields.
- Open the workspace the path routes you through and capture inputs as a URL (shortlist, comparison, brief).
- Run the matching AI Usage Lab playbook + prompt set in your own model harness — the platform does not call live models on your behalf.
- Verify every claim against the citation registry; check the reverification queue for anything due for re-read.
- Export an evidence brief and pair it with your own workload-specific tests before integration.
- Treat every recommendation, ranking, or certification concept as out of scope. The catalogue surfaces evidence; the decision is yours.