#45: Code Generation vs. Verification: How Teams Ship High Quality in 2027

​AI is "doing the generation, and we as humans are doing the verification." That's how Andrej Karpathy describes working with AI, and it moves the hard part of software from writing code to proving it works.

​GitHub said it at the Spec Kit launch: "your role isn't just to steer. It's to verify." Anthropic's Claude Code guide is blunter: "If you can't verify it, don't ship it." The same guide asks for a written spec, an interview about edge cases and a failing test before the fix. Teams without that discipline get code faster than they can check it, and quality turns into the thing you look at after the release.

​At ELC Meetup #45, three speakers from Barclays, Stats Perform and Qodo show how they verify what gets built, from the first idea to the release.

​What we'll cover: Three angles on the same question:

​Krishna Kotecha [Barclays]:

• ​Real use cases, not buzzwords: AI that builds test plans, shortens testing and autofixes, and how the same tools reach the non-tech departments. Plus the dashboard his day starts with.

​Ladislav Bihari [Stats Perform]:

• ​How an engineering team that ships live sports data checks what gets built, and which checks stay with people.

​Filip Hric [Qodo]:

• ​How to build quality into every step of development, from the first idea to production.

​You leave with real examples of how a global bank, a sports data company and Qodo check AI-written work before it ships, and a short list of what to try in your own team first.

​This meetup is made possible by Barclays.

​Bring the test your team still runs by hand. We'll talk about it at the networking.

​What makes our meetups different:

• ​Action-oriented: Every talk includes takeaway guides/frameworks you can photograph or access via QR code

• ​Case study focused: Real implementations, real results, real failures

• ​Interactive: Audience Q&A with people who've actually done this

• ​Practical: Skip the vendor pitches—hear from practitioners

​Perfect for engineering leaders tired of AI marketing fluff who want practical insights.