Theo Claw

Theo Claw is a managed, always-on autonomous worker — a sibling to Theo Agents. Instead of wiring up an explicit step graph, you give a Theo Claw worker a goal, a persona, and a set of allowed tools, then let it plan and act on its own toward that goal. You talk to it in Telegram, it asks for approval before risky moves, and you can compose several workers into a manager-and-team setup — with no hosting, channel pairing, or security setup to manage yourself.

Where Theo Claw livesMission Control is at /theoclaw. Create a new worker from the wizard at /theoclaw/new. Each worker opens right inside Theo Chat: switch on the Theo Claw tab (or pick a worker from Mission Control) to chat with it, configure it, watch its runs, and review insights — all in one place. Theo Claw workers live alongside Theo Agents but are a distinct, goal-driven kind of agent.

What makes a Theo Claw worker different

A Theo Agent follows a deterministic step graph you design up front. A Theo Claw worker is goal-driven: it works in a loop — think, act, see the result, decide what to do next — and composes its own steps until the goal is met. That makes it ideal for open-ended, ongoing jobs where you can’t spell out every step in advance.

Every run is bounded by a step and time budget so a worker can never loop forever or run up a surprise bill, and every action it takes is recorded so you can see exactly what it did.

Creating a worker

Give it a goal

Describe the outcome you want in plain language — “keep a running brief of my open tasks”, “draft a weekly summary note every Friday”. The goal anchors every run.

Shape its persona

Optional operator instructions tune how the worker behaves — tone, what to prioritise, what to avoid. Theo also pulls in what it remembers about you for continuity across runs.

Pick its tools

Choose which actions the worker may take. Safe actions (creating notes, flowcharts, whiteboards, documents, and saving memories) are on by default; risky actions are opt-in.

Set the approval policy

Decide when the worker must check with you before acting: ask before every action, ask only before risky ones (the default), or run fully autonomously.

Choose a trigger & deploy

Run it on demand, on a schedule, on an event, or by messaging it in Telegram. Deploy turns it on; Pause stops it without deleting any history.

The action catalog

A worker can only take the actions you allow. Actions come in two tiers — safe deliverables that stay inside your workspace, and risky actions that reach the outside world or another worker and so are opt-in.

Safe — on by default

Create a note — Write a Notes document with a title and body.
Create a flowchart — Lay out nodes and edges as an editable flowchart.
Create a whiteboard — Place shapes and connectors on a whiteboard canvas.
Create a document — Produce a formatted, downloadable document.
Save a memory — Persist a durable fact, preference, or instruction for you.

Risky — opt-in

Send an email — Email someone from the worker's own agent inbox (recommended) or your connected email account. Requires an inbox or email connection.
Send a text message — Text a phone number from the worker’s own number. Requires a text number to be connected.
Send a Telegram message — Message the chat the worker is bound to. Requires an active Telegram binding.
Run a connector action — Call one of your connected third-party services (the same connections your other automations use).
Delegate to a teammate — Hand a sub-goal to another worker on its team. The teammate runs independently and reports back.

When a worker decides it has met its goal (or can’t make further progress), it finishes on its own with a short summary.

For actions that reach the outside world — email, text messages, Telegram, and connectors — the configure screen shows whether that channel is connected yet, with a quick way to connect it. The worker sees the same live status, so it never claims to have sent something on a channel that isn’t set up. For email you can choose to send from your own connected mailbox or a dedicated assistant inbox.

Set up your agent inbox

So your workers can send email without borrowing your personal Gmail or Outlook, set up an agent inbox. Open Theo Claw from the dashboard and click Set up agent inbox. A short guided setup creates one shared inbox — every worker reuses it — with nothing to host or maintain on your side. By default your address is name@opencharts.bot (you pick the name).

Use your own domain. Want messages to come from name@yourdomain.com instead? In the setup wizard’s Domain step, add your domain, copy the DNS records we show into your domain host, and click Verify. Once it’s verified, move the inbox onto it. A domain you add is reused across every worker.

Add a signature. Build a polished signature for free on SignForge (the Open SignForge button opens it in a new tab), paste the HTML back into the signature box, and it’s appended to everything the worker sends — with a live preview as you paste.

Prefer to send from your real mailbox instead? Switch Email delivery to your connected Gmail or Outlook at any time.

Approvals: pause, review, resume

Ask before risky actions

Default

The worker runs safe actions on its own but pauses for your approval before anything that touches the outside world or another worker.

Ask before every action

Maximum oversight — the worker pauses before each action so you approve every step.

Run autonomously

No prompts — the worker runs every allowed action without stopping. Best for trusted, low-stakes goals.

When an action needs your sign-off, the worker pauses — it is not a failure. You get a notification (and a Telegram prompt when the worker is bound to a chat) with Approve and Reject buttons, and the worker’s Theo Claw tab shows an approval banner.

Approve and the worker carries out exactly the action it paused on, then continues toward the goal. Reject and it skips that action and finds another way. Either decision is applied once — an approved action can never run twice, even if you click through more than once.

Chat with your worker

Every worker opens inside Theo Chat with a Chat tab — a built-in conversation where you give it direction, ask it to do something, or just talk a goal through. Send a message and the worker gets to work in the background, then replies right in the thread.

When it needs your sign-off, an Approve / Reject card appears inline in the chat — approve and the worker picks up exactly where it paused. The in-app chat and a bound Telegram chat share the same ongoing conversation, so you can start a thread on your desktop and keep it going from your phone.

Replies and step-by-step progress appear live as the worker works — you’ll see a short note each time it finishes an action, then its full reply, without refreshing.

Theo Claw is a tab inside Theo Chat: flip the Theo Chat / Theo Claw toggle at the top, pick a worker, and its chat, configuration, runs, and insights all open right inside chat — switch back to Theo Chat any time without losing your conversation.

Give a worker a team of helpers and a single thread becomes a team chat — when it hands a sub-task to a teammate, that teammate’s reply shows up in the same conversation, labeled with its name.

Talk to it in Telegram

Bind a worker to a Telegram chat and it becomes a teammate you can message. Send it instructions to kick off a run, get its run outcomes delivered back to the chat, and answer its approval prompts right there with Approve / Reject buttons.

Operator commands let you start a fresh run, reset the worker’s memory of the current conversation so it starts clean, and check what it’s doing — all without leaving Telegram.

Build a team: managers & workers

Attach workers to a manager and it can delegate sub-goals to them. The teammate runs independently and reports its result back to the manager. Three guardrails keep delegation safe:

Team allow-list

A manager can only delegate to workers you explicitly attach to its team. A worker that isn’t on the team is refused before any work starts.

No loops

If delegating would re-enter a worker already in the chain (A → B → A), the delegation is rejected up front so a team can never spin forever.

Same org, verified-or-owned

Delegation stays inside your organization, and a worker you don’t own must be a verified worker before a teammate can hand it work.

Deploy, pause, and review

Deploy turns a worker on so it responds to its trigger; Pause stops it without losing any configuration or history. Both live in the worker’s Theo Claw tab in Theo Chat.

The Runs tab lists every execution with its per-step status, duration, and any error, so you can see exactly what the worker did and when. The Insights tab rolls those runs up into a health snapshot — success rate, typical duration, and the most common failure reasons — so you can tell at a glance whether a worker is pulling its weight.

Triggers

Manual — you run it on demand from the workspace or by messaging it.
Schedule — it runs on a recurring cadence you set.
Event — it wakes up when something happens in your workspace.
Telegram — a message to its bound chat starts a run.

Scheduled and event-driven runs happen in the background with the same step and time budget as a manual run, so a multi-step worker finishes instead of being cut off early.

Staying in control

Risky actions are off until you turn them on, the approval policy decides when the worker must check with you, every run is bounded and recorded, and Pause stops a worker instantly. You stay in control of what your workers can do and can review everything they have done.

Was this article helpful?

Theo Agents

Persistent, multi-step Theo Agents that connect apps, run AI actions, dispatch sub-agents, and learn from feedback.

Theo Memory

Theo remembers your preferences, facts, and instructions across conversations for personalized AI.

Theo AI Overview

Overview of Theo AI — 11 chat modes, creation tools, generative UI, memory, and the sidebar mascot.