Your agents don’t need to persist. Your purpose does.
How do I connect Asana to ChatGPT?
The Work Graph over MCP serves any client, so ChatGPT can ask Asana what exists, who owns it, and how it rolls up to goals.
Direct answer: connect Asana's MCP server to ChatGPT. In ChatGPT, add it from the connector settings (developer mode); the tool's own docs carry the current path. Once authorised, ChatGPT reads the Work Graph — tasks, projects and goals with their connections — so planning questions are answered from live structure rather than from memory.
The graph answers from either side
Due-this-week across teams, the project's open blockers, how the quarter's goal is fed by its projects — one server, answered identically wherever you ask. The goals layer keeps the answers attached to intent.
Closure is a checklist you promised
Worked case: your team set a handover rule — no client project closes without its handover task complete, priced by a project that ended twice: once when the work finished and again a month later when the client discovered nothing had been handed over. Asana tracks the tasks; that closure waits on one of them is your promise about closure.
Through Unl the project read arrives with the promise held: marked done, handover task still open — not closed by your own rule, the twice-ended project being the why. The rule survives staff changes because it carries its story.
Asana to ChatGPT reads the Work Graph; through Unl it reads the graph against the closure promises you set — 'done' means what you defined, not what the status field says.
Reads through Unl arrive with measured context — in the presence of the decisions you’ve already settled. It works today: one box, paste anything. If it speaks MCP, Unl can reach it. Readings arrive unprompted, the data beside the criterion; Unl is a courier, not a warehouse, and keeps only your keys and the frame.
Read further
Questions people ask
What is the Asana read surface over MCP?
Tasks, projects and goals with their connections — the Work Graph as structured data, same across every MCP client.
Is there anything ChatGPT-specific to configure?
Only the client-side connector setup. The server, authorisation and surface are the same as for any other MCP client.
What closure rules are worth deciding?
Whatever 'done' keeps failing to mean: handover complete, docs updated, invoice sent. Each agreed definition turns a status read into a verification.
What if work is marked done without meeting your definition?
Then the read shows you the gap between the status field and the standard, which is the point of holding the standard somewhere the field cannot edit. What it will not do is close the gap for you. The rule surfaces the disagreement; the conversation is still yours to have.
What this is
Your agents don’t need to persist. Your purpose does.
Save the thoughts, decisions and targets worth keeping, each with its reasoning, carried into every AI session the moment they matter. Unprompted.
Your whole AI world. What you decided at the epicentre. It works inside Claude, Claude Code, ChatGPT and Cursor, and connects in a couple of steps.
MCP native·You decide·Model agnostic·Your data
The right information
Connect Unl to bring the right information into the moment.
Your sources, read against the criteria you set.
The full product, open. Free to start. Paid plans from $10 a month.