TownGiving your assistant context
Your assistant knows you - your voice, your priorities, your key contacts. But so far, it can only see your email, calendar, and Drive. That is a good start. It is not the full picture.
Imagine your new team member only has access to your email. They can help you draft replies and triage your inbox, but that is where it stops. If you ask them to reschedule a meeting, they cannot - they cannot see your calendar. If you ask them to check on a deal, they cannot - they do not have CRM access.
Now imagine you give them access to your calendar, your CRM, your project tracker, and your team’s Slack. Suddenly they can do a lot more. Not because they got smarter, but because they have more information to work with.
This is exactly how integrations work in Town. Your assistant already knows you. Each tool you connect gives it more to know about.
Context is the whole game
This is worth stating directly: your assistant is only as good as the information it has access to.
It already knows how to search, read, summarize, draft, and act. What it needs is information. Every tool you connect is another source of truth it can draw on - another dimension of your work it can see and act on.
Think about it this way. If someone emails you asking about a deal, your assistant can draft a reply in your voice, but only if it can look up the deal in HubSpot. If you need to prep for a meeting, your assistant can pull context on the attendees, but only if it can see who is on the calendar invite and check relevant Slack threads.
The more your assistant knows about your work, the more it can do for you. Same capabilities, richer context, dramatically better output.
A real example
Rather than listing what each integration does, let us walk through a scenario.
You get an email from a customer asking about the status of their project. Here is what your assistant can do depending on what is connected:
- With just email: It can draft a reply in your voice, but it has to ask you for the project status because it cannot look it up anywhere.
- Add calendar: Now it knows you have a check-in with this customer next Tuesday and can reference that in the reply.
- Add Slack: Now it can find the latest discussion about this customer’s project in your team channels and pull in the relevant context.
- Add HubSpot: Now it can pull up the customer’s account, see their deal stage, and reference notes from your last conversation.
- Add Linear or GitHub: Now it can look up the actual project status - open issues, recent commits, what shipped last week - and include specifics.
Same email, same task, dramatically different quality of response. Each connection did not add a new capability - it added knowledge about your work. And the draft still sounds like you - it just has more substance behind it.
Start with what matters
The mechanics of connecting are simple. Go to Integrations, click Connect, sign in. Done.
The real question is: what should you connect first?
Start with what is relevant to your daily work. If you live in Slack, connect Slack. If you manage deals, connect your CRM. If you work with code, connect GitHub.
You do not need to connect everything on day one. Start with what matters most, see how your assistant uses the additional context, and expand from there. You will notice the difference immediately - the suggestions get more specific, the drafts get richer, and the workflows can do more.
Your assistant will tell you what is missing
Remember the integration suggestions from Chapter 1? As your assistant works with you, it notices when it is missing context it needs.
If it sees you reference Slack conversations in emails but cannot search Slack directly, it will suggest connecting Slack. If it is trying to prep you for a meeting but cannot look up the attendee’s company in your CRM, it will flag that.
These suggestions are your assistant saying “I could help you more with this if I could see that.” It is worth paying attention to them.
Custom connections
What if the tool you need is not on the list?
Town supports MCP servers, which let you connect custom tools - internal APIs, niche services, databases. If it has an API, you can probably connect it.
One thing to know: when you connect an MCP server, your assistant gets access to everything that server exposes. There is no way to toggle individual actions on or off yet. So review what the server provides before connecting, and use approval-required mode until you are confident in the behavior.
Same principle as our delegation analogy. You would not give your new hire admin access to a critical system without understanding what they could do with it first.
A common mistake
People sometimes connect a tool and expect their assistant to immediately start using it in every workflow.
That is not quite how it works. Connecting a tool makes it available, but you still need to enable it on each workflow that should use it. Think of it like giving your team member login credentials - they have access, but you still need to tell them which projects to use it on.
If you connect Slack but your morning briefing does not mention Slack activity, check whether the Slack tools are enabled on that workflow.
What’s next
Your assistant now knows you and has access to your tools. That is the foundation. In the next chapter, we will put it to work - teaching your assistant to handle things automatically, without you asking every time. That is where workflows come in.