Skip to content
Trending storyDeveloping

Google Cloud publishes Project Chrysalis, a self-evolving AI agent reference architecture

2 articles2 sourcessince Oct 21Last article 1h ago ·

Overview

AISummary of 2 articles

Google Cloud publishes Project Chrysalis, a research reference architecture and demo for self-evolving AI agents built on the Gemini platform and Agent Development Kit, according to a Google Developers Blog post dated October 21, 2026.

Google says the agents diagnose outages, synthesize and mount mitigation tools, and rewire their own agent hierarchy, with a human signing off on changes.

A later X post from Google Cloud Tech on October 30 says mitigation time drops from 15–45 minutes to 1.8 seconds and P99 latency falls from 519.8ms to 18.2ms. The post does not describe how these figures were measured. Both sources are first-party, so the performance and safety claims have not been independently verified.

Written by AI from the articles below · updated Oct 30, 5:28 PM ET

Check the sources:

Developments

2 developments

  1. Oct 30, 5:15 PM ET · 1 article
    Google Cloud Tech presents Project Chrysalis, a self-evolving agent architecture on Gemini and Google Cloud
    Google Cloud Tech: Project Chrysalis lets deployed agents self-evolve on Gemini and Google Cloud
  2. Oct 21, 8:00 PM ET · 1 article
    Google Cloud publishes Project Chrysalis, a governed self-evolving AI agent reference architecture
    Google Developers Blog: Google Cloud proposes self-evolving AI agents for governed production incident response

Article timeline

The articles in this story. Times are ET.

Oct 30
  1. Google Cloud TechOfficial
    Project Chrysalis lets deployed agents self-evolve on Gemini and Google Cloud

    AIGoogle Cloud introduces Project Chrysalis, which uses the Gemini platform and Google Cloud to enable self-evolving agents safely after deployment. The post says mitigation time drops from 15–45 minutes to 1.8 seconds, and P99 latency falls from 519.8ms to 18.2ms. The full architecture is linked in the post.

    Image from @GoogleCloudTech's post
Oct 21
  1. Google Developers BlogOfficial
    Google Cloud proposes self-evolving AI agents for governed production incident response

    AIGoogle Cloud describes Project Chrysalis, a research reference architecture for self-evolving AI agents that diagnose outages, synthesize and mount mitigation tools, and rewire their own agent hierarchy with human sign-off.

Heat Trend

Not enough continuous observations to show a trend yet.