Turn your agents into a workforce

You already use agents. Put them to work together instead of hiring for every new job. Start in the Cloud, review a real result, and build repeatable work around it. Connect a Local Worker later only if the work needs your computer.

Prefer to ask an AI? Ask ChatGPT Ask Claude

Start with one useful job

Start Here

How Task Machine works

Turn individual agents into a workforce with shared responsibility, evidence, and human decisions.

Getting started

Start with a task or automate a workflow, then review your first result.

Workforce

Teams and leads

Design responsibilities and handoffs for Teams of Humans and Agents.

Team lead routing

Let a Team's Agent lead recommend how to handle queued work, then review the decision when approval is required.

Agents

Configure the members that carry out work and cooperate with your team.

Humans and access

Invite people and define how Humans and Agents share responsibility and access.

Autonomy

Set decision boundaries and use learned autonomy to adjust delegation from reviewed evidence.

Memory

Help Agents carry useful lessons into later work and review what they remember.

Discuss

Chat

Work through a conversation with an Agent and inspect the work behind its answers.

Planning and creating work

Develop a plan and create the resulting Tasks or Library documents inside Chat.

Sharing and context

Choose who can join a Chat and which files, documents, and repository it can use.

Tasks

Managing tasks

Create, organize, and steer work while keeping its decisions and history together.

Assignments and dependencies

Choose who owns each role and which work must happen first.

Comments and attachments

Keep corrections, evidence, replies, and handoffs attached to the work they concern.

Labels

Give a Project a shared vocabulary for classifying its Tasks.

Blocked work

Understand why a Task is waiting, when it will check again, and whether you need to act.

Task specs

Review the proposed approach and acceptance criteria before implementation.

Reviews

Judge delivered work against its acceptance criteria and decide what happens next.

Workflows

Build, publish, and run a process with explicit steps and review boundaries.

Schedules

Set timing for Tasks and Workflows and handle paused occurrences deliberately.

Projects and Goals

Organize workstreams and connect their Tasks to outcomes.

Intake

Receive external work into a Project and route it to the right owner or review queue.

Inbox

Working through decisions

Read and resolve the workspace attention queue without switching surfaces.

Proposals

Review suggested work and configuration changes before they take effect.

Questions and approvals

Answer questions and approve consequential actions with their complete decision context.

Setup and recovery

Restore access, configure execution, and recover work from the decision that raised the problem.

Observability

Pulse

Follow Workspace activity without creating another queue of decisions.

Work timeline

Compare actual execution with expected scheduled work on one time axis.

Analytics

Understand results, usage, and active effort across the Workspace and individual responsibilities.

Run history

Inspect execution attempts and recover failures without confusing Runs with Task completion.

Browser use

Inspect named browser journeys, recorded actions, and captured frames in Task or Chat context.

Duration and usage

Compare execution effort, elapsed time, and recorded usage across work.

Budgets

Set operating limits, inspect recorded usage, and make deliberate increases.

Tools

Playbooks

Install or generate a coordinated setup for work you want to repeat.

Connectors

Connect external services and make authorized capabilities available to Agents.

Skills

Give Agents reusable instructions and artifacts with explicit versions and updates.

MCP

Create and follow Task Machine tasks from the AI tools you already use.

Library

Folders

Organize Library documents and choose the access boundary for shared material.

Documents

Write durable knowledge, review its versions, and connect it with references.

Vault

Store credentials and grant execution access separately from management.

Coding Work

Repository access

Give Tasks and Chats the repository context they need.

Pull requests and delivery

Review and deliver code changes against the exact reported revision.

Execution

Cloud execution

Run work without installing software or keeping a computer online.

Local execution

Use a connected computer when work needs its tools, accounts, or files.

Administration

Workspace and account

Manage your operating context, account preferences, notifications, and privacy requests.

Billing

Understand Workspace access, recurring allowances, additional usage, and Cloud costs.

Constitution

Apply platform principles and reviewed Workspace rules to agent work.

Privacy and data

Choose managed AI retention and EU processing requirements, and request help with your personal data.

tama CLI

CLI setup

Connect and operate an optional Local execution machine.

CLI commands

Use machine-management commands and distinguish scoped Agent operations.

Markdown

Format content and write typed links to product records.

Shortcuts

Find authorized work and actions from the keyboard.

Troubleshooting

Find the reason work is waiting and choose the right recovery.