Set up an agent, step by step
Why this exists
An agent is useful only when it works from the right facts: your customers, your prices, your rules and the sources it may read. If it guesses, its drafts are wrong and nobody trusts them. So Sasha builds each agent with you in one chat. It checks what it already knows, asks only about the gaps, shows you the brief before it saves it, and runs once for you to check before anything goes on a schedule.
An example
The CRM Agent at Context is Everything works from its CRM Brief. Each day it reads the customer list, a day's activity, the open requests, the fixes that shipped in the last 7 days and the help articles. Its drafts go from the Send as address in section 6 of the brief. It runs every day at 07:00 London time. An admin can change each of these on the brief tab.
The set-up starts from the panel that each role shows before it is set up. It says what the agent needs, and what Sasha checks first.

How to do it
The steps below are in order, from the first connection to the first scheduled report. You need an admin account.
Before you start: give Sasha enough context
An agent is only as good as what Sasha knows about your business. Each agent lists what it needs before it is built: for example, the Content Creator Agent needs how you write, with two or three real pieces, and at least five meetings or project notes from the last month. You can see the list in the Set it up box in Agents that work for me.
When you start a set-up, Sasha first checks each item against what it really holds: your business brain, your projects and your connected tools. It then says Ready, Thin or Not ready, with the evidence for each item.
- Ready: the set-up goes on.
- Thin or Not ready: Sasha says what the agent would get wrong, and gives you one prompt for each gap. Paste each prompt into a new chat. Sasha asks you a few questions and writes the answer where every agent can use it. Then ask again to set up the agent.
You can also tell Sasha to go on without filling the gaps. Sasha then asks those questions during the set-up, and the agent starts with less to work from.
Before you start: connect the sources
An agent can read only what is connected to Sasha. Connect each source first, so that Sasha can find it during the set-up. Each agent's page says which sources it needs. For the CRM Agent, see the CRM Agent page.
There are three kinds of source.
| Source | Where you add it | Use it for |
|---|---|---|
| A server for a service | Settings, Tools, Add MCP Server | Live data from a system, for example your customer list or your cloud bill |
| A GitHub synced folder | Settings, Shared Drives, Add synced folder | Documents that your team writes in GitHub, for example help articles |
| GitHub's read-only server | Settings, Tools, Add MCP Server, Use a key | The history of your code: what shipped, and when |
A server for a service: Sign in or Use a key
The Add MCP Server form has tabs. Use the tab that the service tells you to use. See Connect third-party tools to Sasha for the full steps.
- Sign in. You sign in on the service's own page. Sasha keeps that sign-in for everyone. The agent can read only while the server stays signed in. If the server shows "Needs an admin to sign in again", the agent cannot read it until an admin signs in again.
- Use a key. The service gives you an address and a key. Put the key in Headers, for example
Authorization=Bearer your-key. Use an equals sign, not a colon.
Choose the Server Name with care. You type it exactly in a later step, and a spelling mistake in the name is part of the name. Do not change the name after an agent uses it.
After you save a server, press Test. A working server shows its tools, for example "12 tools available." Note the names of the tools that only read. You need them later.
A GitHub synced folder for documents
A synced folder copies one folder of a GitHub repository into a Sasha project, and refreshes it every day. The agent reads the copy as part of that project. See Sync a folder from GitHub into a project.
GitHub's read-only server for code history
A synced folder copies files. It does not copy the history. To let an agent see what shipped and when, add GitHub's own read-only server with Use a key. The steps are in Connect third-party tools to Sasha, under "Read a GitHub repository's history".
Step 1: start from Agents that work for me
- Open Agents that work for me in the side menu.
- Click the agent with Set up next to its name, for example CRM Agent. A panel says what the agent does and what it needs.
- Click Set it up. Sasha starts a new chat, labelled Setting up: CRM Agent above the message box, and puts the request, for example "Set up our CRM Agent.", in the box.
- Press Send.
A new chat keeps the set-up apart from your other work. When Sasha creates the agent in that chat, the chat moves under the agent in the left menu by itself, so it is the first chat you find when you open the agent. See Find your chats again, and choose where a new chat lives.

Step 2: answer Sasha's questions
Sasha reads what it already knows about your business and your connected sources. It tells you what it can see, what it cannot read yet, and what is probably missing. Then it asks only about the gaps, a few questions at a time.
Answer in your own words. When Sasha proposes a rule from your real data, check it. For example, Sasha can read Northwind's customer list and see a group that has access but does not pay. It asks what that group is before it makes a "trial" stage.
Sasha finds each source itself. It does not trust a name that you give it. If your help articles are in a different project from the one you named, Sasha says so and uses the correct project.
Step 3: check the brief
Sasha writes the agent's brief: the job, the rules, who gets the report and when. It shows you the brief before it saves it. Read it and correct what is wrong. Later, an admin can change the brief on its tab in the app, or ask Sasha in chat.
On the tab, each section is one row that says what is in it. To change one setting, type its name in Find a setting, for example Send as, and change it where it appears. An admin sees the fields straight away, and nothing changes until they press Save, either beside the changed field or in the bar. There is no undo after saving, so each changed field shows what it was, with Put it back.
The brief has settings, a reading list and notes. The agent reads the settings and the reading list on every run. The agent does not read the notes. Notes are for the people who read the brief. A rule that the agent must follow goes in a setting or in the reading list, never only in the notes.
Step 4: choose what the agent may read
Sasha builds the agent and turns on the read tools that it can. It cannot always save these two things, so an admin does them:
- the read tools of a server that you added yourself under Add MCP Server;
- the knowledge projects that the agent may read.
Sasha tells you this, and gives you the exact lines to enter. Then:
- Open the agent under Agents that work for me.
- Open the ⋯ menu and choose What it may read.
- Tick each read tool that is offered. If a tool is not offered as a tick box, type it in Tools it may use to read your figures, one per line. Write each tool as
mcp__<server>__<tool>, for examplemcp__northwind-customers__listCustomers. The server name must match the name in Settings exactly. - Enter read tools only. Never enter a tool that writes, sends or deletes.
- Under Knowledge it may read, tick each project that the agent needs, for example the project that holds your synced help articles. A project that is not ticked is not read. Naming a project in the brief does not give access.
- Press Save. A Confirm box opens in the Studio window. It lists what the agent may read, what it may change (usually nothing), the knowledge projects and the limits. Check them, then press Confirm.
For a picture of what an agent reads and of the Confirm dialog, see the product guide page on team roles.
Step 5: run the first report
On the Report tab, press Run now. You can also ask Sasha in the set-up chat to start the run. Sasha then asks in the chat, for example "Start a run of the CRM Agent now?", and says: "It takes a few minutes. When it ends, Sasha may email the report to the people on its list." Sasha starts no run until you press Start it. Not now starts nothing.
If no read tool is saved yet, the run says "set up needed" and lists what to connect. That is correct. Go back to step 4.

Step 6: review the first report
Read the first report with the person who owns the job.
- Compare each figure with a figure that you trust.
- Look at "What it could not read". A source in that list is "not read". It is never taken as zero or as all well.
- Correct the brief where it is wrong, then press Run now again.
An admin can give the report a grade and write questions in Review, in the ⋯ menu. The next run answers each question.
Step 7: put it on a schedule
Do this only after you approve a test run. Ask Sasha in the set-up chat to put the agent on its schedule, for example "Run it every day at 06:00". The schedule then shows in Menu, Tasks, where an admin can switch it off.
Work through the drafts
A CRM Agent also writes drafts. Each one sits on the Report tab with Send, Open and edit, the marks and Bin. See Approve, change or bin what an agent drafted for each button. The CRM Agent page has a picture of each step:
- the draft card, unmarked, with its row of buttons;
- the question that Send asks in the card;
- a sent draft, closed to one line;
- the edit page, with the history of every change;
- Bin it on the draft's page, with Never write to them ticked.
Who can do this
- Admin: sets up an agent, saves what it may read, saves the brief, and puts it on a schedule.
- Staff: reads the reports and works through the drafts. Cannot set up an agent or save the brief.
- Member: cannot sign in to the Sasha web interface.
If something goes wrong
| What you see | What to do |
|---|---|
| The set-up chat stopped in the middle, for example on a memory limit | Send "continue from step N" in the same chat. N is the last step that Sasha finished. Sasha keeps what it already wrote. |
| The report shows a source as "not read" | Open What it may read. Check that each read tool of that source is listed and saved. Check that the server name in each line matches the name in Settings exactly. Press Test on the server in Settings, Tools. |
| A Sign in server stopped working | The server must stay signed in. In Settings, Tools, find the server. If it says "Needs an admin to sign in again", an admin clicks Sign in again. |
| The agent does not follow a rule | Check that the rule is in a setting or in the reading list of the brief, not only in the notes. |
| The agent cannot read a project | Tick the project under Knowledge it may read, then Save and Confirm. |
More detail
- What an agent does each day: Let an agent do a regular job for you.
- Schedules: Run a job on a schedule, without asking each time.
- Reports, briefs and settings: the product guide page on team roles.
- Apps in general: Use and manage apps built on your knowledge.
Questions
Why can the agent not read a server that I connected in Settings?
Connecting a server does not give an agent access to it. An admin must also enter the server's read tools in the agent's What it may read, then press Save and Confirm.
Sasha said that it could not turn on the tools or tick the projects. Is that an error?
No. Sasha cannot always save the read tools of a server that you added yourself, or the knowledge projects. An admin does these two things in What it may read. Sasha gives you the exact lines to enter.
I wrote a rule in the notes of the brief. Why does the agent not follow it?
The agent does not read the notes when it runs. Notes are for people. Put a rule in a setting of the brief or in its reading list.
The set-up chat stopped in the middle. Do I start again?
No. Send "continue from step N" in the same chat, where N is the last step that Sasha finished. Sasha keeps what it already wrote.
Sasha says it does not know enough about my business to build the agent. What do I do?
Paste each prompt that Sasha gave you into a new chat and answer its questions. Sasha writes the answers where every agent can use them. Then ask again to set up the agent. You can also tell Sasha to go on now and answer the questions during the set-up.
Can an agent send a message or change something?
Not on its own. An agent reads, writes a report and makes drafts. A person can edit or bin a draft. Sasha sends only what a named person approved, in the exact form they approved, at the time they approved, and only while that person may still send. A person presses Send, approves the draft for a day and time on the Schedule tab, or sends it from their own email. Any other action waits in Held items until an admin approves it.
What do the buttons on a draft do?
Each draft on the CRM Agent's Report tab has one row of buttons. Send sends the draft from your address, after one question in the card. If sending is not set up, it shows a Copy button instead, so you can paste the message into your own email and send it. Open and edit opens the draft so you can change it. Sent myself, Not sent and They replied tell the agent what happened to a draft you dealt with yourself, and the next run counts your mark. Bin, on the right, throws the draft away. The agent itself cannot send or approve anything. On the Schedule tab you can also approve a draft for a day and time, and Sasha sends it then, exactly as you approved it. The CRM Agent page on the docs site has a picture of it.
Can Sasha send the draft for me?
Yes, once it is set up. Press Send on the draft. The card asks "Send this to" the person, and says which address it goes from and that replies come to you. Press Send now and Sasha sends it, from the Send as address in the CRM Brief (section 6, Who follows up). The card then says when it went and from which address, and the next run counts the person as written to. Sasha sends a draft only when a named person sent it with Send or approved it for a day and time on the Schedule tab, and never to anyone on the never-write list. Send as must be an address your Sasha administrator has set up for sending. Until it is, Send offers to copy the message instead. You can also ask Sasha in chat, for example "send the note to Ben". The CRM Agent page on the docs site has a picture of it.
How do I edit a draft before I send it?
On the CRM Agent's Report tab, click Open and edit on the draft. Change the message, and the To or Subject if you need to, then press Save. Sasha keeps every version, with who changed it, when and from where. You can also ask Sasha in chat, for example "make the note to Ben shorter". An admin or a staff user can edit a draft. The CRM Agent page on the docs site has a picture of it.
How do I bin a draft, and never write to that person again?
On the Report tab, click Bin on the right of the draft. It is binned at once, with the reason Not needed. To give a different reason, or to add the person to the brief's never-write list, open the draft with Open and edit, choose why under Bin it, tick Never write to them if you want to, then press Bin. Only an admin can add someone to that list. For 30 days the draft stays on the Report tab, greyed, with Undo to bring it back. You can also ask Sasha in chat to bin one draft or many. Sasha asks once, and bins nothing until you press the button that says yes. The CRM Agent page on the docs site has a picture of it.
Can I open a draft from the report email?
Yes. Under Drafts waiting, each draft has a line that starts Open it, followed by the link. The link opens the agent in Sasha at that draft, where you can edit it, mark it or bin it.