The same thing, inside the product
Everything in this module, without a key or a config file — because the client's own team gets the same capability inside the product.
Everything so far in this module has been your tools reaching into a client's org. That is the partner's half. The client's own team gets the same capability from inside the product, and it is worth thirty seconds of a demo because it is the part they will actually use.
Same capabilities, a panel attached
Xen is the conversation, inside Expedify. It is not a second product and it is not a different assistant — it reaches the same capabilities you have been calling through MCP, over the same org. What it adds is a workspace panel: ask for something and the thing you made opens beside the conversation, live, rather than as a link you have to go and find.
What it looks like
One instruction — the same dashboard you built across four asks in Building a dashboard by asking — and the dashboard opens in the panel next to the chat. Same three objects underneath, same org, same numbers you checked on 2026-08-19.
That panel is the difference, and it is why the client's team will use Xen where a partner uses MCP. Nobody in a branch office wants a config file. They want to ask for something and see it.
Which one is for whom
| MCP | Xen | |
|---|---|---|
| Who | You, in your own tools | The client's team, in the product |
| Set-up | A key and a config block | Nothing — they are signed in |
| What you see | The answer | The answer, and the thing it made |
| Best at | Working across several client orgs, scripting, building things repeatedly | One org, in the moment, by somebody who does not want a terminal |
Who
- MCP
- You, in your own tools
- Xen
- The client's team, in the product
Set-up
- MCP
- A key and a config block
- Xen
- Nothing — they are signed in
What you see
- MCP
- The answer
- Xen
- The answer, and the thing it made
Best at
- MCP
- Working across several client orgs, scripting, building things repeatedly
- Xen
- One org, in the moment, by somebody who does not want a terminal
They are the same capability catalog with different collars on. Anything this module taught you to ask for through MCP, a client can ask for in Xen — and anything they build there is an ordinary workflow, chart or dashboard that opens on the canvas exactly as everything in Modules 3 and 4 did.
The rule for both, and it is the module's point
Conversation for the thing you do once. Canvas for the thing that runs forever.
- Ask when the work is a one-off. A question before a meeting, a bulk edit nobody wants to click through, a first draft of something you will then read properly.
- Open the canvas when it will run without you. Anything that fires on a trigger, anything a client's team must maintain, anything where a wrong arm costs somebody a customer.
- Never activate from a conversation without looking. Everything you built in this module came back a draft on purpose. Draft and active: which workflows are actually running is still the lesson.
A partner who gets this right looks fast and is safe. One who builds everything by asking and never opens the canvas is fast for about three weeks.
Change one thing — ask for the same thing twice
- Open Xen in your demo org and ask it for a dashboard, in one sentence.
- Watch the panel rather than the reply. That is the thing to show a client.
- Now open what it made on the ordinary dashboards screen. It is just a dashboard.
- Delete it, and decide which of the two surfaces you would have used for that job.
Try it
- Work out which member of a client's team you would show Xen to first, and what you would have them ask.
- Decide what you would say to a client who wants their whole team building workflows by chat.
That is the conversation module. You can reach a client's org from your own tools, answer questions nobody built a screen for, change records, and build a workflow, an agent and a dashboard by describing them — and you know when not to. Next: The twelve-minute demo.
Related lessons
Base rates — what a piece of evidence is actually worth
A face-recognition system that is 99.9% accurate and almost entirely wrong, and a number that sent an innocent woman to prison. Both are the same arithmetic, and it is the arithmetic that decides what any piece of evidence is worth.
ReadConfirmation and survivorship — what you never looked for
Two questions about evidence you did not go looking for. One is a rule you have to discover, and one is a pattern in five famous people — and in both, the thing that would have told you the truth is the thing nobody checks.
ReadLoss aversion, sunk cost and regression — what it costs you
Four questions you answer about yourself rather than about a scenario, and your own answers are the finding. Then the pattern that makes praise look useless and criticism look like it works, whatever you actually do.
Read