How AI Agents Saved Upsales From Crashing In Production

Upsales is a CRM system that helps sales teams work smarter. It handles tasks ranging from mass mailing to managing deal flows. Their platform is central to daily operations, which means even a small bug can lead to major disruptions.

Company: B2B SaaS

Dev team: 20 devs

Test cases: 70

Before Automation

  • The QA process was slow as it was all manual.
  • Bug reproduction required watching long videos created by users.
  • Smoke tests were manually triggered.

After Automation

  • Automated QA reduces time to deploy and covers edge cases.
  • Bug reproduction is now easy to handle.
  • Daily smoke tests are automated.

The Problem: Manual Testing Wasn’t Covering Enough

Upsales initially relied on manual QA, which was time-consumin,g and it often missed real user flows that led to failures.

Their team needed a way to quickly test common CRM usage scenarios without writing complex scripts. On top of it all, QA testers were watching videos of the reproduced bugs, which wasted hours of their time.

The Solution: QA.tech and AI Agents

QA.tech’s AI agent was introduced to simulate and test real human interactions in the CRM system. It can easily navigate through and use the CRM like a real user, uncovering critical bugs without manual setup.

“Smoke tests in Beta now run every day instead of manually once a week.”

This way, two crash-causing bugs were discovered by QA.tech. One of those was a mass mailing issue that didn’t reach any users.

The Results: Downtime Prevention

The Upsales team now uses the QA agent to continuously test high-traffic parts of the CRM, helping catch issues early.

Instead of burdening the QA team with loads of videos, bug reproduction feature is used, which makes it easy to handle and create reproductions.

Overall, both the developers and QA team can now rely on the agent to handle repetitive test tasks, which enables them to focus on tasks that require human intervention.

“We’re really impressed by the speed of development from the QA.tech team.”