SAP
AI-Powered SaaS for B2B Developers, 2025
Challenge & Solution
Automated tests are often flaky and break as applications change, forcing testers to spend too much time maintaining scripts instead of testing software. SAP runs hundreds of thousands of quality insurance (QA) tests annually with millions dollar cost.
To speed up the test efficiency, our solution automatically detects failures, uses browser agents to verify the intended behavior, and heals brittle Playwright tests so they remain reliable as the UI evolves.
My Role
Product Design Lead: I led the end-to-end design of a next-gen AI product, from defining the problem and planning research to validating hypotheses, improving models with AI scientists, and delivering pixel-perfect designs.
Impact
Integrated to SAP Self-Healing Test Automation with potential to scale across 100,000+ tests
Team
1 PM, 1 Designer, 2 SWE, 2 AI Scientists, 2 Product Owners
Worked closely with SAP AI Lab through UnternehmerTUM, DPS
Timeline & Status
3 months
Launched 1st MVP in May 2025
User
We're Building for Technical Experts
A Engineers & Automation Leads
DevOps & SRE Teams
Problem
“95% of automated tests fail due to annoying flakiness”
Automated tests fail all the time. Testers spend more time fixing test scripts than testing code. “Test Flakiness” is a huge problem. It occurs when a test fails unpredictably but was expected to pass.
Solution
ZeroFlake: Our tool uses the power of AI to eliminate test flakiness.
Our solution automatically detects and fixes broken test scripts. We run each Playwright test, capture failures, and use browser agents to verify the intended business logic. The agents convert brittle Playwright steps into more resilient AI-driven tests, then heal the original script so it continues to work reliably as the UI changes.
Process
Vibe Code, Prototype, Testing, Validating Hypotheses, and Coffee
Yeah, I was the only designer on the team, so I wore a lot of hats. I did everything a designer and researcher would normally do, and even took on some of my product manager’s work. I worked closely with scientists to run experiments, testing different factors and input/output parameters to validate which ideas worked and which didn’t.
Final Delivery
Step1: Upload Test Script
Step 2: Run the Test Script
Step 3: Get AI Suggestions
Step 4: Update Test Script
Step 5: Re-run the Test
Impact
Integrated to SAP Self-Healing Test Automation with potential to scale across 100,000+ tests
The final result was highly praised by SAP’s Head of Ecosystem Development and members of the AI Lab, and was subsequently integrated into SAP’s AI-driven testing system here.
Testimonial
I found Jeter to be one of the best designers that I have had the chance to work with over the course of my PM career. Jeter was very collaborative, very open throughout the design process, very thoughtful/creative, and he was incredibly fast-paced and receptive to feedback.
It must also be mentioned that Jeter brought a lot of humor and positivity and warmth to the team, which helped our team bonding and overall atmosphere. I can only hope to partner with more Designers like Jeter in my future projects!
Credits
Designer: Jeter
Product Manager: Alex
AI Scientists: Karun, Vidushee
Back-end Engineer: Yang
Front-end Engineer: Sara
Consultant: Marcus
Product Owners: Leonard, Benedikt
Ecosystem Development: Tobias
Special thanks to UnternehmerTUM for their support