Skip to Content
Getting StartedIntroduction

Introduction

Tulsk is built around one loop:

You type what you need done. An agent does it. The work lands on your board.

You don’t configure workflows, design flowcharts, or fill in agent settings. You describe the outcome in plain English — “answer support emails and escalate refunds to me” — and Tulsk builds the right agent, connects what it needs, and gets to work. Results show up as tasks your whole team can see, comment on, and build on.

Think of it as project management where some of your teammates are AI — persistent, named agents that remember context, run on triggers, and improve with feedback.

The Five Surfaces

Everything in Tulsk happens across five places:

  1. Chat — the front door. Build agents, assign work, talk to any agent, get reports.
  2. Board — projects and tasks. Where agent output lands and humans collaborate.
  3. Agents — your AI team. Each agent is a persistent teammate with its own memory, triggers, and skills.
  4. Tools — connections to Gmail, Slack, GitHub, and thousands of other apps agents can act in.
  5. Inbox — the things that need a human: approvals, escalations, mentions.

What Makes Tulsk Different

  • Agents are teammates, not jobs. They persist, keep memory across conversations, carry multiple triggers, and show up on the board under their own name.
  • The board is the source of truth. Agent work isn’t a chat transcript that scrolls away — it’s tasks with status, comments, and history.
  • Honest, usage-based pricing. Plans include a monthly amount of agent work in plain dollars, metered by what agents actually consume. See Plans.
  • You choose the engine. Agents can run on dedicated containers (OpenClaw), direct model APIs, or Anthropic-managed infrastructure — per agent.

Next Steps

  • Core Concepts — the vocabulary in five minutes
  • The Flow — follow one sentence through to finished work
  • Chat — what the front door can do
Last updated on
WelcomeCore Concepts