Hi Google Antigravity team,
I am a developer building an algorithmic trading bot on Google’s stack using
Cloud Run, Firestore, BigQuery, Secret Manager, and Gemini 3.1 Pro as the
AI decision layer.
While building this system, I noticed a significant gap between what
Antigravity currently offers and what Claude Co-work provides on the
Anthropic side.
Claude Co-work gives users a single, opinionated “agent operator” experience
where an AI can sit alongside your project, follow a constitution of rules,
take scheduled actions, manage files, monitor workflows, and act as a
persistent co-pilot — all from within a polished desktop-aware interface.
Google Antigravity is incredibly powerful for building agentic apps, but
right now that power is spread across AI Studio, Vertex AI, Cloud Run, and
Workspace tools. There is no single “turn it on and it works with you”
experience that a solo developer or power user can activate the way
Claude Co-work can be.
My feature request is:
-
A Co-work style persistent personal agent layer inside Antigravity, where
a user can define a set of rules, tools, and goals, and the agent
operates continuously or on schedule without requiring manual prompting
each session. -
A desktop-installable PWA or native app shell for Antigravity that
functions like a control room — showing agent status, task history,
decisions made, and live monitoring — so it feels like a real product
rather than a developer playground. -
Native Secret Manager and Cloud Run integration within the Antigravity
UI, so keys, webhooks, and deployment configs can be managed in one
place alongside the agent logic.
This would make Antigravity competitive with Claude Co-work for serious
personal and developer use cases, and would make Google the strongest
end-to-end agentic stack available — which it nearly already is, just
missing this cohesive operator layer.
Thank you for building Antigravity. It is a genuinely exciting platform
and I hope this feedback is useful.