ai·rete·rag
Paste a policy doc, get an auditable decision engine
5 followers
Paste a policy doc, get an auditable decision engine
5 followers
The AI drafts a rule graph from your policy, then steps out of the decision. A deterministic engine decides every verdict; the AI only explains it, grounded in your docs. Same facts, same answer — with the exact rules that fired. No guessed verdicts.
Free Options
Launch tags:SaaS
Launch Team
Render: Get up to $100K in startup creditsThe fastest way to deploy and scale apps and agents
Promoted
It would be great if the tool could highlight which specific rule fired and show the exact clause from my docs that triggered the verdict, so I can quickly defend decisions during compliance audits without digging through the explanation.
@enayy8ya Yes that is possible at
@enayy8ya
This is exactly the workflow I built it for, and most of it ships today:
Which rule fired — every decision shows the exact rules that fired, in causal order. For rules that didn't fire, it shows which condition failed and the value it actually saw (e.g. credit_score >= 720 (got 650)). So "why this verdict, and why not the other one" is answered on the decision itself, not buried in the prose.
Source evidence — each decision also returns the policy excerpts it retrieved, with the source document and relevance, shown next to the verdict.
The one piece not yet wired end-to-end is exactly what you're pointing at: when you import a policy document, each generated rule is shown with the exact sentence it encodes — but that citation currently lives in the import review step, not on the saved rule. Carrying it through so it sits next to the fired rule in an audit is a small change and clearly the right one. I'm adding it.
Genuinely the sharpest feedback I've gotten today — if you run compliance audits day to day, I'd love to hear what else you end up digging for.
@enayy8ya
Update — this is live. Every rule now carries the policy sentence it encodes, shown as a "Source policy" block right on the rule. In an audit you see the fired rule, the conditions and actual values that matched, and the clause it came from, side by side. Rules created by policy import get this filled in automatically; you can also add or edit it by hand. Thanks for the push — it was the right call.
Rule view:
Audit:
@enayy8ya
did citation end up working for your use case, or did you hit anything confusing?
Saw the same refund case get the same verdict twice and the rule trace actually showed why, which is rare. The AI summary being separate from the decision is a smart way to keep things auditable.