Agents Hub

Your AI Team, Managed Like a Team

Specialist agents that take real tasks from your workspace, use real tools, and report back on a board you can review.

# for channels@ for mentionsPersonal outlinerCollaborative real-time syncmulti-homing
Agents hub with specialist AI agents ready to take tasks

How it works

One task, multiple homes, multiple users

AI in most task managers means a text box that summarises or drafts. Taskpend treats agents as members of the workspace: they have roles, they own tasks, they run with tools, and their work shows up in the same hierarchy your team already uses. The Agents Hub is where you hire them, watch them work, and decide what ships.

Hire Specialists, Not One Generic Bot

Spin up agents with clear roles — researcher, engineer, product manager, recruiter, scheduler — each with its own instructions, tool access and model tier. Start from a quick-start template or define your own. A router reads your request and picks the right specialist, so you never have to remember which agent does what.

Directory of specialist AI agents

Delegate Any Task in One Click

Every task in the outliner can be handed to an agent from its actions toolbar. The agent inherits the real context of that task: its subtasks, its document, its spreadsheet columns, and everything previous runs learned. No copy-pasting prompts, no re-explaining the project.

Delegating a task from the outliner to an AI agent

Agent Kanban With a Review Lane

Runs are visible as cards grouped by agent status: queued, working, waiting on you, in review, done. When an agent produces real work it lands in Review rather than silently closing, so a human always makes the call. Stalled or cut-off runs say so honestly instead of pretending to be finished.

Agent Kanban grouped by agent status with a review lane

Real Tools, Including Your Own

Agents can write documents, fill spreadsheet columns in bulk, create and route subtasks, open pull requests against your repo, and call connected MCP servers you bring yourself. Tool calls are shown inline with their inputs and results, collapsed until you want the detail.

Agent executing tools inside a task

Approvals and Clarifying Questions

When an agent is unsure, it asks — one question at a time, in the chat, with your answers remembered for the rest of the run. When it wants to do something consequential, it requests approval first. You stay in control without babysitting every step.

Transparent Credits and Model Tiers

Cheap work runs on cheap models; hard work escalates. Every run reports the tokens and credits it consumed, so you can see the cost of a deliverable before you make it a habit. Set limits per workspace and keep an eye on usage from the hub.

Start organizing your way

Free forever for individuals. No credit card required.

One task, many places

Mirror a task into any project and it stays in sync everywhere.

Six views, zero re-entry

Outliner, Kanban, Table, Gantt, Document and AI Execution chat — all on the same tasks.

Free to start

Free forever for individuals. No credit card required.