AI code assistants with Peter Guagenti

A newsletter that started as a personal learning and getting-in-touch project by Ivan Krnic grew into a sociotechnical newsletter covering topics of technical excellence, organizational improvements, and productivity.

AI is moving into every segment of software delivery and maintenance. AI code assistants help us write new and understand legacy code. In API management, AI identifies cyberattacks. In AIOps it predicts when things could go wrong.

These are just some of the examples of how AI can help. But every organization is different. What helped us the most in finding the right opportunities was throwing an internal AI hackathon where we spent a day tinkering with the technology but keeping in mind the challenges that our clients and we face daily. For CROZ, as a BizTech consultancy, AI is about finding ways to bring our clients more value and make our work more efficient. Some of the best ideas came straight from this AI hackathon – check out my DevOps Enterprise Summit talk.

Still, I can’t help but wonder…

When we got pocket calculators, many of us forgot how to divide numbers by hand…
When we got automatic transmission, many of us forgot how to drive a stick…
Now AI code assistants are here. Will we forget how to code? And should we fight it or embrace it as a sign of an evolution?

This is just one of the topics I discussed with Peter Guagenti, the President of Tabnine. In Peter’s eyes, AI code assistants should be Iron Man suits, making each developer better, stronger, and more efficient. AI code assistants are nowhere near replacing us for complex stuff. As my friend Damir said, today’s AI is a basic reasoning engine. To unconditionally trust code assistants like we trust pocket calculators and automatic transmissions, we need to reach the level of an advanced reasoning engine.

The way the industry has been moving for the last two years, anything is possible.

We’re back in Zadar for another QED conference and a bonus conference-within-conference Mighty Mainframe!

Our keynote speakers will be Sam Newman, technologist and best-selling book author, and Rosalind Radcliffe, IBM Fellow, CIO DevSecOps CTO.

Register here and join us!

Tech predictions for 2024 and beyond – Werner Vogels predictions for 2024, check out the part about AI assistants redefining developer productivity.

New GitHub Copilot Research Finds ‘Downward Pressure on Code Quality’ – some studies warn that using code assistants is correlated with “mistake code” being pushed to the repo, less code moving, and more copy-pasting. All this suggests less refactoring according to the DRY principle and more quick code generation without thinking of system sustainability. It’s another warning that code assistants, like any other tool, can produce value if used correctly but also damage if misused.

langchain4j – using LLMs from Java has never been easier.

How does Pixar consistently create hit after hit? How did Hendrix build his studio? And why was the construction of Heathrow Terminal 5 an incredible success while the construction of Wembley flopped?

There are patterns in successful large projects, and this book by Prof. Bent Flyvbjerg and Dan Gardner will reveal them all and set you up for success.