Karim Juini built an ERP for a cyber cafe in high school, later co-founded Expensia (acquired for $100M+), and now leads Thunders to revolutionize software testing with AI. The discussion explores how AI-driven QA can replace slow, costly traditional methods, enabling faster innovation and strategic quality roles like “quality architect.”

Into The MoTaverse

Rosie Sherry interviews people involved in testing. Video only interviews. Available on youtube or the homepage. Each episode has a full transcript if you find it on the main site.

• https://www.ministryoftesting.com/podcasts/into-the-motaverse
• https://www.youtube.com/playlist?list=PLbdLjg29s9lCY4hspzj3AGdAL7Vr2ys1B

Episode Details

• Show Notes: https://www.youtube.com/watch?v=KKdjrxL_sx0
• Published: 2026-09-04T13:58:04Z
• Duration: 00:50:58
• Author: MoTaverse

Overview

Karim Juini shares his journey in tech and entrepreneurship, beginning with early coding experiences and launching a mini ERP software for a cyber cafe during high school, sold via a subscription-like model. He later worked at Microsoft for 7 - 8 years, where he developed a passion for software testing and scaling systems. He co-founded Expensia, an AI-driven expense management platform that scaled to 2 million users across 60 countries before being acquired for over $100 million. Two years ago, he founded Thunders to focus on improving software quality, particularly in testing, support, and scalability.

The discussion highlights the evolution of software development and quality assurance over the past two decades, emphasizing the growing complexity in testing, especially for AI-powered products operating under strict regulatory environments. Traditional QA methods are seen as costly and slow, leading to innovation bottlenecks. Generative AI is presented as a transformative solution - combining the scalability of automation with the adaptability of human-like testing. The future of QA is envisioned as more integrated and strategic, with roles like “quality architect” emerging to guide quality across development cycles, while AI handles execution and regression testing at scale.

What If

• What if you delegated 80% of your manual testing to AI agents this quarter?

  • Move: Identify your top 3 repetitive, high-maintenance test cases (e.g., regression, login flows, form validations) and rebuild them using an AI-powered testing tool like testRigor or Applitools. Run them in parallel with your existing suite for 2 weeks to compare coverage and false positives.
  • Why Now?: AI testing tools now support natural language inputs and self-healing locators, eliminating the maintenance burden of Playwright/Selenium scripts. With release cycles accelerating, manual or script-heavy testing is becoming a bottleneck.
  • Expected Upside: Free up 15 - 20 hours/month for higher-value work (e.g., edge-case design, UX validation), reduce production bugs by catching regressions earlier, and shorten release cycles from weekly to daily.

• What if you became the “Quality Architect” in your solo dev workflow starting next week?

  • Move: Map your current development loop (idea code test deploy) and insert quality checkpoints: (1) define acceptance criteria in plain English before coding, (2) use AI to generate test scenarios from those criteria, (3) run automated visual + functional checks post-build, and (4) log results in a lightweight dashboard (e.g., Notion or Airtable).
  • Why Now?: The shift from siloed QA to integrated quality ownership is accelerating. As a solo operator, you’re already all roles - this formalizes quality as a proactive function, not a final gate. Tools now exist to automate execution while you focus on strategy.
  • Expected Upside: Reduce post-launch fixes by 40 - 60%, improve user retention through consistent quality, and create a repeatable system that scales when you eventually hire or outsource.

• What if you allocated 20% of your weekly dev time to AI-driven test delegation instead of writing test scripts?

  • Move: Block 1.5 hours daily (or 7.5 hours weekly) to explore and implement AI testing tools - spend Week 1 experimenting with tools like GitHub Copilot for test generation, Week 2 integrating an AI tester into your CI/CD pipeline, Week 3 measuring defect escape rate pre/post, and Week 4 refining prompts and coverage.
  • Why Now?: The “waitism” trap - waiting for the perfect AI tool - is costly. The core skills (prompting, validation, integration) are transferable across tools. Starting now builds muscle memory before AI becomes the default expectation.
  • Expected Upside: Cut test creation time by 70%, increase test coverage (especially edge cases), and position yourself as an early adopter - giving you a competitive edge in delivering reliable software faster than peers relying on legacy testing methods.

Takeaway

• Audit your current testing process and identify one repetitive task to automate or delegate using AI tools within the next week.
• Dedicate 20% of your weekly work time (e.g., 1 - 2 hours) to learning and experimenting with AI-powered testing tools or platforms.
• Reframe your role to include quality ownership by integrating basic QA checks into your development workflow before handing off features.
• Start migrating legacy test scripts incrementally using AI assistance instead of rewriting them from scratch or maintaining outdated automation.
• Focus on system thinking by mapping how your software interacts with external systems, regulations, and user behaviors to anticipate scalability and compliance issues early.