How much AI development has already changed

This week's must-know news for engineering leaders

Author

Ada
April 22, 2025

This week’s bytes:

  • Exactly how AI development has already changed

  • Who should engineering managers trust?

  • A GitHub security disaster

Worth reading

🤖 Beyond chatbots: what actually good AI development looks like

7-minute read by Loraine Lawson

AI agents aren’t chatbots, and automation isn’t about prompts. Fractional AI shares why engineering workflows—not just prompting models—is the real unlock. From handling hallucinations to creating evals, these lessons show what serious AI development actually looks like.

4-minute read by Kelli Korducki

A growing number of engineering managers are turning to generative AI as their go-to thinking partner. From strategy to feedback, AI offers judgment-free, always-on support. But relying too much on machines may cost leaders the human insight their teams actually need.

3-minute read by Matt Foster

Over 20,000 GitHub repos were exposed after a malicious update to a popular Action. This supply chain breach reveals a chilling truth: most teams trust third-party Actions blindly. Are your CI/CD pipelines silently vulnerable? Here's what went wrong and how to lock yours down.

2-minute read by Julie Bort

Ex-Y Combinator president Geoff Ralston is betting big on “safe AI” with a new fund. While others chase speed and power, he’s backing startups that prioritize safety, security, compliance and guardrails in AI deployment.

5-minute read by Alex Ponomarev

From floating-concrete feature requests to contradictory demands, engineering managers constantly juggle the unrealistic. But instead of saying no, great EMs decode the real need, reframe the ask, and guide teams toward sane solutions. Magic? Maybe. But it’s more strategy than spellwork.

Best quick bytes

#1. 🏆

#2. 🏆

#3. 🏆

Ready to be more productive as a team?

You’re going to love our signature reports: Impact, Flow, DevEx, and Snapshot.