Customer story
A risk decisioning platform where every request touches fraud, identity and compliance logic. Customer feedback now moves from support ticket to reviewed pull request without leaving the team's existing GitHub workflow.
The challenge
Feature requests come through support, sales calls, integration partners and developer questions about the API. The pattern behind them was never in one place.
The platform spans document checks, biometrics, eIDV, AML screening, device and payment signals. Every request competes with the next layer of the roadmap.
Changes near verification and audit paths need a written rationale and a review, not a fast merge.
What shipped
How it runs
Support tickets, calls and developer questions group into themes automatically.
AutomaticThemes are scored by revenue at stake; a person approves what becomes a spec.
Their decisionA PRD is generated, then an agent writes the change and runs the suite in a sandbox.
Cost tracked liveThe change opens as a normal pull request with the originating tickets cited.
Their decisionScoped to the repos ScreenlyyID selects, with write limited to branches and pull requests.
The tickets or calls behind a change are linked in the description for reviewers.
Every job's spend is visible as it runs — no black box.
A ScreenlyyID engineer approves every pull request.
With a small engineering team, our backlog used to be a queue that would shift based on the most recent customer call. Small fixes would wait weeks behind big features. Now it moves on evidence and clarity on what's holding us back. By documenting everything, and generating architectural decisions that can be reviewed by a senior engineer before implementing, the amount of work we can run in parallel is incredible.
The parts of VocxAI in ScreenlyyID's day-to-day.
Run this on your own roadmap.
In private beta — join now and we'll map it to your workflow.
Join private beta