Run a policy evidence check
Compare three fictional candidate answers against an explicit reference. See which source checks fail.
Run the example →
RESOURCE LIBRARY
Useful material. Open to inspect.
Start with the sample evidence, prepare a project brief or explore the engineering guides. Downloads are free and do not require an email address.
NO SIGN-UP REQUIRED
Choose what helps your next decision.
Read the findings and limitations
A small, complete demonstration report with the method, results and reproducible inputs.
Download report →Structure a real evaluation
An unfilled template for scope, coverage, run conditions, uncertainties and engineering recommendations.
Download template →Inspect the reference and candidates
Fictional current and retired policies, plus the exact fields used by the browser and Python checks.
Download fixture →Reproduce the checks locally
The same deterministic checks outside the browser. No model API, credentials or network connection required.
Download runner →Prepare the first conversation
A short checklist for your workflow, decision, existing assets and access constraints. No private records needed.
Download brief →Plan a complete evaluation
An eight-step playbook covering task selection, evidence, grading, reproducibility, failure analysis and handoff.
Download playbook →Review the release decision
Evidence-linked checks for quality, runtime operations, rollback, ownership and remaining risks.
Download checklist →
USE THE LIBRARY
From the first brief to the release review.
-
01
Describe the workflow
Use the project brief to identify the task, decision, failures and constraints.
-
02
Plan the evidence
Follow the playbook to select references, task slices and grading rules. Inspect the fictional example to see how separate evidence checks work.
-
03
Record and review
Use the report template during the run, then review the release checklist with the accountable owner. Templates need evidence from your actual system.
READ FURTHER
From the connected engineering practice.
Acadify engineering blog
Explore implementation guides →
Read about AI products, evaluation and practical engineering decisions.
Our approachUnderstand the wider evaluation scope →
See how references, test conditions and human review fit together.
Public examples demonstrate an approach. They do not represent client outcomes or production model accuracy.
START A CONVERSATION
Tell us what your AI needs to get right.
Share the workflow and the problem. We’ll confirm fit, then discuss scope, access, deliverables and cost before work begins.
Discuss your project