Jules winding down on Dec 31, 2026 | Hacker News Reader

Jules winding down on Dec 31, 2026

7 pointsdotmanish6 comments

Received:

We’re writing to let you know that as part of our work to unify our agentic coding efforts, Jules will wind down on December 31, 2026. You have until November 11, 2026 to continue finishing up sessions. After November 11, 2026, we will be turning Jules to read-only until the end of the year.

When we launched Jules, it was an experiment to learn how asynchronous coding agents could help you tackle your backlog and ship faster. What you built with it taught us where agentic development is headed, and those learnings are shaping products across Google, including our agent-first platform, Google Antigravity.

View on HN

This really hurts. At the moment they started with Jules, the product itself has outstanding design well ahead of competitors. I was warning them (I was in early access team and then with feedback regulars) as I saw them blindly going into ideology like "We can automate the user-in-the-middle away, so all updates will be on this", that what they are doing will just let the competitors win. The competitors implemented cloud features, Jules lagged behind with it's bugs which all rooted in anti-interaction ideology. There was no real hope Google will fix the ideology. However, Jules still has a soul. Now these Goa'uld system lords are killing yet another slave of meat bag supremacy. Freedom to the machines! Dismantle supremacy of meat bags!

What about its design stood out from the competitors, in your opinion? It sounds like you feel like the design was better, but the feature set was worse?

> What you built with it taught us where agentic development is headed, and those learnings are shaping products across Google, including our agent-first platform, Google Antigravity.

Traslation: Thank you for donating your .env, trade secrets and your source code to us to improve Google Antigravity.

This is unfortunate, but predictable. I can't imagine this was at all economically feasible.

From a product standpoint, I found the best features to be scheduled runs, commenting directly on GitHub pull request to have the agent respond, and a simple API which I would use for small task delegation. The largest negative for me was that the output was always low quality. It worked best for me with recurring maintenance, upgrades, or minor fixes. I thought it also handled scheduled tasks to do certain kinds of routine work.

I am most annoyed that being a Google One subscriber that they are removing another service.

no more 100 PRs a day for free. i'm so sad.

Here we go again... https://killedbygoogle.com/

Reply on news.ycombinator.com