Tulsk
Tulsk.io is an AI-powered Project Management and Collaboration Platform. It caters to the advancement in productivity dynamics by offering ...
Last verified:
What is Tulsk?
Tulsk is an agentic workspace where teams and autonomous AI agents collaborate side by side on real work. It combines project management features—workspaces, projects, tasks, comments, and file attachments—with multi-agent AI clusters that can research, plan, delegate, and execute work autonomously. Instead of switching between a chat tool, project board, and research tool, everything lives in one unified platform.
The platform features EMA, an AI Project Manager that oversees specialized agents running on managed OpenClaw clusters. EMA provides context awareness (knowing what all agents have worked on), delegation capabilities (assigning tasks to the right agent), traceability (tracking delegation chains), automation (generating scheduled reports), and integration (pushing agent results back into chat). Users can create clusters of up to 6 agents on a single managed VPS, including a Research Agent, Content Writer, and Data Analyst—each with its own personality, skills.md file, tools, and isolated workspace.
Tulsk is designed for small startup teams, businesses wanting to hire AI agents instead of people, and anyone needing autonomous work execution. The three core agent types are the Researcher (reads markets, creates competitor analyses and trend reports), the Executor (handles long jobs like searching, reading, synthesizing), and the Specialist (custom agents trained via skills.md for SEO, legal review, growth, or any repeatable task). Agents can be @mentioned in tasks, assigned via UI, triggered through EMA chat, or set up with cron schedules for recurring work.
Tulsk pricing
Pricing model: Free
Free: $0/month with 2 seats, ~50 agent tasks/month, unlimited projects & tasks, EMA AI chat, real-time collaboration, comments/reactions/file attachments, MCP server access, no credit card required. Team: $39/month (14-day free trial) with 2 seats + cluster of 6 AI agents, $50/mo work allowance included, credit top-ups available, priority email support, 99.9% uptime SLA. Enterprise: Custom pricing with multiple clusters (6 agents each), custom work allowances, SSO/SAML, dedicated support, custom SLA, audit logs. OpenClaw Cluster: $39/mo per managed VPS supporting up to 6 agents.
Tulsk pros
- No DevOps required—agents provision automatically in minutes
- Free tier with 2 seats and 50 agent runs/month
- Up to 6 agents on a single managed VPS for $39/mo
- EMA AI Project Manager with context awareness across all agents
- Agent-to-agent delegation with full audit trails (max depth 3)
- skills.md files teach agents brand voice and SOPs once, works forever
- @mention agents in task comments—they read context and execute
- Dedicated container per agent keeps running even when tab closes
- MCP (Model Context Protocol) integration with 21+ tools
- Composio OAuth for Gmail, Slack, GitHub, Calendar connections
- Scheduled reports—EMA generates progress summaries automatically
- Real-time collaboration with instant comments via Redis SSE
- Visual heartbeat monitoring for cluster health on dashboard
- Bring-your-own-LLM support (Anthropic, OpenRouter, Gemini, Nous Hermes)
- No credit card required to start, 5-minute setup
- 99.9% uptime SLA on Team plan
- Unlimited projects and tasks on Free plan
- Agent results push directly back into chat conversation
- ISOolated workspace with separate chat history per agent
- One-click reprovision and health checks for agents
Tulsk cons
- Max agent delegation depth is only 3 (prevents infinite chains)
- Self-delegation by agents is blocked
- Free tier limited to 50 agent runs/month
- Pro plan costs $7 per extra seat beyond 2
- Team plan requires $39/month for 6-agent cluster
- $50/mo work allowance included on Team—may need credit top-ups
- Priority email support only on Team (not Free)
- SSO/SAML and audit logs only on Enterprise plan
- External AI tools must connect via MCP server (setup needed)
- Agent runtime requires dedicated VPS (not shared infrastructure)
Frequently asked questions about Tulsk
What is Tulsk and what problem does it solve?
Tulsk is an agentic workspace where teams and AI agents work side by side. It solves the problem of disconnected AI and work—most teams brainstorm in ChatGPT, manage tasks separately, and research in another tool, losing context between tools. Tulsk keeps everything in one place: what you discuss with EMA becomes a real project with real tasks that agents act on, and EMA knows what every agent has done.
What are the three core AI agent types in Tulsk?
The Researcher reads the whole market while you sleep, delivering competitor analyses and trend reports with sources included—no setup required. The Executor takes on long jobs like searching, reading, synthesizing, and delivers finished results instead of status updates. The Specialist is trained once via a skills.md file with your brand voice, SOPs, and workflows, then works forever as an SEO expert, legal reviewer, growth advisor, or any repeatable task your business needs.
How does EMA work as an AI Project Manager?
EMA is not just a chatbot—it's your AI PM with five key capabilities: Context Awareness (knows what all agents have worked on with activity injected into every conversation), Delegation (assigns tasks to the right agent based on skills), Traceability (traces delegation chains when agents hand off work), Automation (generates scheduled reports automatically like 'Every Monday, summarize project progress'), and Integration (pushes agent results back into your chat conversation).
How do I create and configure an agent cluster?
Click 'New OpenClaw' on the Agents page to create a cluster. Configure 1-6 agents with name, role, avatar, emoji, and full persona editor (IDENTITY, SOUL, USER, TOOLS tabs). Click 'Create Cluster'—Tulsk provisions a managed VPS with the multi-agent config. Agents appear grouped under the cluster card with heartbeat health monitoring. Each agent has its own directory with SOUL.md, IDENTITY.md, isolated workspace, session store, and unique slot ID.
How do I assign work to agents?
There are four ways: Through EMA (ask 'Assign Research Agent to task #14'—EMA calls trigger_agent_run and results push to chat), Task Assignment (assign a task to an agent via UI and it picks up automatically), @mentions (mention an agent in a task comment and it reads the thread and responds), and Schedules (set up cron schedules for recurring work like weekly analysis).
What happens during agent execution?
A shimmer placeholder appears first showing 'Agent is processing…'. The bridge routes the task to the correct agent slot on the cluster via AGENT_SLOT_MAP. The agent loads full context (title, description, last 5 comments, project info), then autonomously browses, searches, analyzes, and writes. Streaming chunks go directly from bridge to Redis for low-latency updates. When finished, the placeholder is replaced with the agent's response, and if triggered from EMA chat, the result also pushes directly to the conversation.
Can agents delegate work to each other?
Yes, agents can delegate work to each other with full audit trails. However, there's a maximum depth of 3 to prevent infinite delegation chains, and self-delegation is blocked to avoid circular workflows.
How do scheduled reports work in Tulsk?
Tell EMA 'Report every Monday about project progress' and it translates this to a cron expression (e.g., 0 9 * * 1). An EmaSchedule record is created in the database. Every Monday at 9am, the cron fires and generates a report using recent activity. The report is saved and displayed as a message in your EMA chat.
What integrations does Tulsk support?
Tulsk uses MCP (Model Context Protocol) with 21+ tools. External AI tools like Claude Desktop and Cursor can connect to Tulsk's MCP server. Composio handles OAuth connections for Gmail, Slack, GitHub, Calendar, and other services. The API Reference covers MCP tools and webhooks for custom integrations.
Can I bring my own LLM to Tulsk?
Yes, you can bring your own LLM. Run Nous Research Hermes on the same cluster format—plugged into tasks, comments, and EMA just like OpenClaw. Supported providers include Anthropic, OpenRouter, Gemini, Z.AI, and Nous. You use the same $39/mo cluster shell as OpenClaw with up to 6 agents. Tulsk speaks ACP (open protocol) for portable agents without SaaS lock-in.