AI Leader · Solution Architect · TPM

Data · Readiness

AI Data / RAG Readiness Scorecard

Don't buy models before data contracts, quality, and retrieval hygiene exist. Score corpus access, retrieval, freshness, ACLs, and ops — get a readiness level plus a 90-day remediation plan.

Free RAG readiness assessment and AI data readiness scorecard. Ten questions on retrieval readiness — not a generic org AI quiz — with in-browser scoring and a remediation plan.

Built for Data · Platform · AI Eng · CISO partners

  • Answers stored in your browser until you optionally email a copy
  • Plain-language result you can paste into a board or CFO brief
  • Directional guidance — validate with a scoped pilot before budget lock

How it works

  1. Work through one question at a time — pick the option that matches how things work today, not your wish list.
  2. Get an instant maturity level (1–5) plus dimension scores.
  3. Share results with risk, platform, or transformation leads; unlock Print / Save as PDF with a work email when ready.

What this tool is for

Teams often buy models and launch RAG demos before data contracts, quality controls, retrieval hygiene, and document-level permissions exist. The result is confident-sounding wrong answers, security leaks, and stalled production rollouts.

Who it’s for: Data platform leads, AI engineers, CISO partners, and transformation sponsors preparing a RAG or knowledge-agent investment — anyone who needs an honest readiness signal before indexing enterprise content.

What you get

  • Overall RAG readiness level (1–5) with plain-language meaning
  • Dimension scores across data access, retrieval, quality, security, and ops
  • A 90-day remediation plan matched to your level — not a generic checklist

AI Assessment Engine analyzes answers with transparent scoring heuristics in your browser — no data leaves the device for scoring.

How scoring works

Each answer scores 1–5. Dimension scores average within that theme. Overall level = rounded average across all questions (Level 1–5).

  1. 1
    Lower scores mark gaps for the 90-day plan.
  2. 2
    Dimension bars show whether risk concentrates (e.g. security vs strategy).
  3. 3
    Planning aid only — not a certification or regulatory attestation.
  4. 4
    Score today's reality (access, ACLs, freshness, evals) — not the roadmap slide.
  5. 5
    Security and quality gaps often outweigh model choice; fix those before scaling corpus size.
  6. 6
    Use the 90-day plan as a shared remediation backlog for data and platform owners.

AI Assessment Engine analyzes answers with transparent scoring heuristics in your browser — no data leaves the device for scoring.

Key terms

RAG
Retrieval-augmented generation: the model answers using retrieved documents from your corpus, not only its training data.
Corpus
The set of documents, tickets, wikis, or data products the system is allowed to retrieve from.
Chunking
Splitting source documents into retrieval units; poor chunking hurts precision even with a strong embedding model.
ACL at retrieval
Enforcing who may see each document when the query runs — not only in the source system like SharePoint.
Golden eval set
A fixed set of representative queries with expected documents or answers used to catch retrieval regressions.

FAQ

What is a RAG readiness assessment?

It scores whether your data access, retrieval architecture, quality, security, and operations are ready for production retrieval-augmented generation — before you invest in indexing and agents.

How is this different from a generic AI readiness quiz?

This scorecard focuses on the data and retrieval stack (corpus access, ACLs, freshness, evals, runbooks). Org-wide AI capability and governance maturity are separate tools on this site.

Should I answer based on our planned RAG architecture?

No. Answer how data and permissions work today. Aspirational scores hide the remediation work that usually blocks production.

What does Level 3 (conditionally ready) mean in practice?

You can run a scoped pilot with defined corpus and users, but retrieval quality, freshness, or ACL gaps should be closed before expanding the audience or corpus.

Does scoring send my answers to a server?

No. The AI Assessment Engine scores in your browser with transparent heuristics. Optional work email only unlocks Print / Save as PDF and follow-up resources.

Can we use this for a board or steering update?

Yes as a directional planning aid with dimension scores and a 90-day plan. It is not a security audit, certification, or regulatory attestation.

What should we do first after a low score?

Follow the level-matched 90-day plan: usually inventory the corpus, name an owner, profile quality and PII, then establish retrieval ACLs and a small golden eval set before heavy indexing.