AI-Driven QA Optimizer: Automated Bug Detection&Improvement

Revolutionizing QA: AI-Powered Bug Detection, Smart Suggestions, Seamless Integration, and Continuous Learning for Optimal Web App Development.

Comment

Who we are

  • Koji - 15 years of industry experience in consumer electronics and AI
  • Olayinka - MBA Grad Berkley | Fellow at Kleiner Perkins
  • Akhil - Berkeley EECS Grad | SWE @ Microsoft
  • Roma - Senior SWE @ SOFI | ex Google

Inspiration

Quality Assurance (QA) testing is critical to ensure that software works as intended before shipping to your customers, but QA is time-consuming and costly to maintain. Existing QA testing apps require a lot of manual work and guidance from developers or other specialists.

What it does

Builds familiarity with your application and creates and updates tests [PoC] [early demo] [market research].

How we built it

GPT-4 is excellent when given clear directions for how it should behave. We leverage GPT-4 to play several distinct roles, Product guide to understand your system, Action agent for writing action scripts, Code critic. The agents work together to write the tests that dive testing for your system.

Challenges we ran into

A robust system for learning and inference with multiple agents has some unique challenges. We leaned on the Voyager paper that recently demonstrated the effectiveness of this approach by creating agents at state-of-the-art levels for open ended tasks.

Accomplishments that we're proud of

Our app can do discover how to interact with you API and run tests based off Natural Language

  • Learned how to use Voyager model with more time
  • Hit milestones that work including (i) API integrations that can be controlled with ChatGPT prompts (ii) Integrating into our code-writing agent

What we learned

AI self-learning agents can make a difference for QA testing ($20 billion market) Lots of great QA test tools can work better with integration AI can have great potential here

What's next for AI-Driven QA Optimizer: Automated Bug Detection&Improvement

-Extend agent functionality to autonomously conduct tests (inc feature coverage) using Voyager model including integrations with testing pipelines -Perform UI System testing with the autonomous agent

Built With

Share this project:

Updates