7 Comments
User's avatar
Augusto's avatar

Did you come across conflicting opinions from seasoned software engineers when writing this article?

For instance, this piece by Glyph [1] to which Armin Ronacher responded to in his blog [2].

[1] https://blog.glyph.im/2025/06/i-think-im-done-thinking-about-genai-for-now.html

[2] https://lucumr.pocoo.org/2025/6/10/genai-criticism/

Expand full comment
Adham Bishr's avatar

Richard Hamming's Book - The Art of Science and Engineering - is a great study for how doing one part of the system better (coding, but now with AI) could produce more solutions on the surface and more problems in the system. If more code is being output, other tools (ex. reviews) have to keep up with the inevitable issues that show up with stronger output from one part of the system.

The real question is does the overall system improve with these tools by a high amount or is it marginal?

Expand full comment
Eric Jeker's avatar

Have some of you guys experimented with Junie from Jetbrains?

https://www.jetbrains.com/junie/

Expand full comment
Dr Milan Milanović's avatar

A nice overview, Gergely. What about Salesforce? Their CEO states that their engineering team is already utilizing AI at a rate of 30-50%, which is significantly better than Microsoft Copilot. This may be investigated.

Expand full comment
Andrew's avatar

Always interesting to see how others in the industry are finding the process, thanks!

A couple of small typos I noticed:

- Marin Fowler: -> Martin Fowler:

- For seasons software engineers -> For seasoned software engineers

Expand full comment
Gergely Orosz's avatar

Thanks for the feedback! All fixed - they slipped through two rounds of editing :-/

Expand full comment
Hilary's avatar

Your link to Warp also has a typo - the link has it "Warpi"

Expand full comment