Other cookies are configurable.
Case study · Fintech
Finverse: an AI agent that tests every release of a fintech platform
We trained an AI agent on the work of Finverse's QA team. It runs unit, API and regression tests, writes documentation for new features and tells QA what to check by hand.
- Client
- Finverse
- Industry
- Fintech · payments
- Service
- AI agent for QA
- Tests
- Unit · API · regression
- Result
- ~60% less manual work
The client
Who Finverse is
Finverse is a fintech platform that handles payments and financial data for businesses and financial services.
The product grows fast: every release brings new features.
The challenge
New features every release, tests that couldn't keep up
Manual regression took up to three days and slowed down every release. Some critical bugs still reached production.
Classic automated tests weren't enough: every new feature meant rewriting them or adding more.
The solution
What the agent learns from, what it produces
We gathered the existing QA cases and a year of the QA team's work into a knowledge base and trained the agent on it. For every new feature the agent writes the documentation, runs the tests and prepares a list of what needs checking by hand. When it finds a bug, it opens a ticket in the team's tracker right away and, once the developer fixes it, re-runs the test and closes the ticket.
Learns from
Produces
The release
Before and after the agent
Before
3 days
of manual regression every release
- ✕The team re-checks almost everything by hand
- ✕Automated tests have to be rewritten for every new feature
- ✕Documentation written by hand, or missing
After
1 hour
of regression, run by the agent
- ✓The agent runs unit, API and regression tests
- ✓QA tests only the new feature and runs a smoke test
- ✓Bugs opened and re-tested automatically
Results
Less manual work, stable releases
"With the new QA process we release with confidence every two weeks. The core financial flows are stable: no regressions, no surprises."
Product Owner, Finverse
Is your QA team chasing releases?
We build AI agents trained on your team's work: tests, documentation and checks, with a fixed price in writing before we start.
All case studies →Related Projects
More case studies from our portfolio.
AI assistant for Google Workspace
An assistant that finds information in Gmail and Drive and drafts replies.
Read the case →NordiskRegnskab: database migration
Moving an accounting platform from a single server to a distributed cloud.
Read the case →Cybersecurity assessment
A security assessment for a green energy company.
Read the case →Related Services
The services behind projects like this one.
AI Development & Integration
Add AI features into a product you already run or are building.
Learn more →