Setup friction is what kills most AI rollouts: weeks of configuration screens before anything works. Connect Turtle AI Coworker to Claude and skip them — describe what you need, and Claude finds the right template, builds it, wires up your tools, and smoke-tests the result. Every call runs under the same roles, policies and audit trail as the dashboard. We lead with Claude because it is the smoothest, and the same connector works in any MCP client.
One chat, start to finish. Here is what the whole loop looks like.
Tell Claude what you need in plain language.
Claude searches the same template gallery as this site and explains why each match fits.
It installs a live, editable copy into your workspace, in dependency order.
Rename things, change tone, swap models, edit prompts. All by asking.
Run a smoke test with sample input and see the result before anything goes live.
Every one of these is a real MCP tool call under the hood. You just talk.
Claude semantic-searches the gallery of agents, teams, AI employees, tables, and solution packs, and tells you why each match fits and which tools it needs.
You get a live, editable copy in your workspace, then Claude applies your changes on top: names, tone, prompts, models.
Tables, knowledge bases, agents, teams, and employees, created and wired in dependency order from your description.
Claude gives you a secure OAuth link to open in your browser. Once you authorize, the tool is wired in. Claude never completes the authorization for you.
Models, LLM keys, agents, teams, tables, knowledge bases, and AI employees: create, read, update, and delete them from the conversation.
Claude inspects your real workspace and returns what is done, what is still to do, and what needs attention, with the exact next step.
Executes an agent with sample input and shows you the result, so you can verify it works before it touches real data.
Claude is how you set up and change your workspace. The dashboard remains home for everything you watch: results, run history, costs, approvals, and billing.
It also kills the empty-dashboard moment. Your first session starts with a working department, not fifteen configuration screens.
When Claude discovers templates for you, it runs on the same discovery engine as the gallery on this website. Browse on the site, install from Claude, or do both. The results are consistent either way.
Turtle AI Coworker exposes a standard MCP server, so the same connector works in Claude, ChatGPT, Cursor, Codex and other MCP clients. Our walkthrough uses Claude because the setup is the smoothest; the exact steps vary by client. No MCP client? The dashboard path is always available and fully equivalent for setup.
For the security review: every MCP call is authenticated as you, answered from the same role matrix, evaluated by the same policy engine, and recorded in the same audit trail as any dashboard action. AI operating AI, inside the same governance layer — not a side door around it.
One connector URL, about a minute of setup, and your next workspace change is a sentence.
