๐Ÿ”” AI in Tech is on our roadmap, in early design โ€” join the early access list
AI in Tech ยท In development

Catch the drift between what was specified and what actually shipped.

We're extending the same reconciliation engine to engineering delivery โ€” comparing specs, tickets, and approved changes against what actually made it to production.

The problem

Specs and shipped code drift apart โ€” and nobody reconciles them.

A ticket says one thing; the merged PR does another. A spec sets a limit; the code enforces a different one. A config change goes to production with no linked ticket at all. Code review catches some of this โ€” it wasn't built to catch all of it, especially once a system has years of accumulated changes.

Zorvaine's engine reads the spec or ticket and the actual shipped change, and surfaces exactly where they disagree.

Concept mockup โ€” not a live product
Ticket: add rate-limit to /api/uploadโœ“ Shipped as specified
Spec: max file size 25MBโš‘ Code enforces 50MB
PR #482 approved changesโœ“ Confirmed
Config change deployedโš‘ No linked ticket

Illustrative example for design purposes only.

How it would work

The same three-step loop we validated in healthcare.

STEP 01

Ingest the pair

The spec, ticket, or approved change โ€” and the actual shipped code, config, or deployment tied to it.

STEP 02

Compare & flag

The model checks behavior, limits, and scope against what was actually shipped, flagging drift and undocumented changes.

STEP 03

Engineer review

Every flag goes to an engineer or tech lead to confirm. Zorvaine doesn't merge, deploy, or revert anything on its own.

Where this stands today

Early design โ€” not yet a live product.

We want to be upfront: AI in Tech is a roadmap item, not something you can pilot yet. What exists today is the detection approach itself, already validated in our RCM product โ€” 94.0% precision on a comparable "does the record match the source of truth" task. We're now scoping what it takes to apply that to spec-vs-code pairs.

Deep integration with a given team's ticketing system, CI/CD, and code review workflow is not yet built. Nothing here reflects a finished or generally available product.
  • Engineering teams in regulated or compliance-heavy software (fintech, healthtech, and similar)
  • Platform/DevOps teams that need an audit trail between approved changes and what's actually deployed
  • Teams whose specs and documentation quietly go stale relative to the codebase
Who this is being designed for

Engineering teams that already feel spec drift, but check for it by hand.

If your team spends time manually auditing deployed changes against tickets or specs โ€” for compliance, for onboarding, or just to keep documentation honest โ€” we'd like to talk to you while we design this.

Join early access

Help us design this the right way.

We're talking to engineering teams before we build. If spec-vs-shipped drift is a real headache for you, fill out the form and we'll be in touch.

Prefer email? Reach us directly at info@zorvaine.com.

Prefer to call? Reach us at +1 (469) 536-0817.