Rubrot
coming soon

A fleet of always-on AI agents.

Join the waitlist to get notified when it's ready.

the rubrot cockpit

Built for autonomous work.

Rubrot receives the same notifications you do, and can act in your workspace autonomously. It learns your workflows, and asks for permission before it touches anything sensitive.

Agent Inbox

Receives the same notifications you do from the platforms you use.

Kickoff — starts in 15m2m
Re: kickoff notes9m
GUM-198 moved to blocked21m
CI failed on widget-web main34m
@zero in #ops — can you take the standup?1h

Approval Policies

Granularly configure access and approval policies for actions.

linear.updateauto
slack.sendask
github.force_pushnever
github.pushgumloop/widget-api

Self Improving

Accumulates knowledge and builds workflows over time.

looking into how we approach schema design…
skills.searchschema-design.md
memory.recallpostgres conventions
runs.rangeruns #212–219
applying what worked last time.

Rubrot is built on our custom harness.

We applied our experience and knowledge from years of agent engineering as an agency into a headless and minimal agent harness that is purpose built for always-on agents and long horizon projects that span weeks or months.

Single Threaded Orchestration

Always talk to the same entity.

The Agent Inbox

A notification system the agent uses to subscribe to incoming events.

Context Sovereignty

The agent uses tools to manage its context window, and decides when to read inbox items.

Intentional Memory

Intentional solutions for episodic, procedural, working and tacit memory.