# AI Specialist Portfolio Evaluation Matrix

Use this matrix before hiring an AI specialist, AI developer, AI studio or contractor. Score each area from 0 to 3.

## Scoring

| Score | Meaning | Action |
| --- | --- | --- |
| 0 | Missing | Do not treat the claim as evidence |
| 1 | Described | Ask for artifacts or a technical walkthrough |
| 2 | Demonstrated | Review the artifact and assumptions |
| 3 | Operated | Validate fit for your own project |

## Matrix

| Area | Evidence to look for | Score | Notes |
| --- | --- | --- | --- |
| Business problem | Problem, user workflow, decision or process improved | 0-3 | |
| Data boundary | Source types, permissions, privacy and cleanup notes | 0-3 | |
| Architecture | Components, prompts, tools, RAG, agents, integrations | 0-3 | |
| Integration behavior | Read/write actions, rollback, ownership of connected systems | 0-3 | |
| Evaluation | Good, bad and edge-case examples with pass criteria | 0-3 | |
| Deployment proof | Runtime, logs, monitoring, release or handoff artifacts | 0-3 | |
| Maintenance | Owner, update workflow, support process and change control | 0-3 | |
| Relevance | Match to your project type, data and risk level | 0-3 | |

## Interpretation

- Any score 0 in data, architecture, evaluation or deployment means clarify before contract.
- Mostly 1 means the portfolio is claim-heavy; request a walkthrough or audit.
- Mostly 2 means evidence is reviewable; run a scoped technical discussion.
- Several 3 scores mean the portfolio shows operated work; validate fit and constraints.

## Recommended next step

| Result | Next step |
| --- | --- |
| Evidence missing | Reject or request clarification |
| Evidence described | Technical walkthrough |
| Evidence demonstrated | Audit or controlled prototype |
| Evidence operated | Contract discussion with project-specific validation |