Context & Memory
What does my Townie actually remember about me?
Your Townie keeps a single, easy-to-read view of what it knows about you. This is what lets it sound like you, make sensible calls on your behalf, and handle things without you having to repeat yourself.
It remembers three kinds of things:
Who you are
The steady picture of you and your world:
- Your role, your company, and what you do
- How you like to communicate — tone, greetings, and sign-offs
- The key people, projects, and topics that come up most often
This is the background context your Townie draws on so its work feels like it came from you.
Your preferences
The specific instructions and notes you’ve shared over time:
- How you want particular tasks handled, like “always include the weather in my morning briefing”
- Feedback you’ve given on how something should be done differently
- One-off facts worth holding onto for next time
Each time you tell your Townie how you’d like something done, it can save that here so it doesn’t have to ask again.
Contacts & people notes
Notes on the people you interact with, so your Townie can act with the right context:
- Who someone is and how they relate to you — a teammate, an investor, a client, a family member
- Helpful background, like their role, company, and what you’ve worked on together
- The history of your conversations and any details worth remembering for next time
This means your Townie can pick up where you left off with someone, instead of treating every interaction as a fresh start.
How it stays current
This view grows as your Townie learns more about you, getting more accurate the more you work together. You can look at it anytime to see what your Townie knows, and you’re always in control — you can add, change, or remove anything in it whenever you like.
Does it remember context across my work and personal accounts?
Yes. Your Townie can see every account you’ve connected, and it builds context from all of them.
One profile, built from all accounts. To learn who you are, your Townie reads sent emails across your connected accounts and builds one shared profile of your style and preferences. Every routine draws on it, so context picked up in one account makes your Townie more helpful everywhere — without ever mixing your accounts up.
It keeps your accounts straight. Your Townie always knows which account a task belongs to and acts within the right one. It can draw on context from all of them when that genuinely helps, but it keeps work and personal cleanly separated — it won’t reply from the wrong address, mix the wrong details, or blur the two.
- Every account is labeled. Each gets a clear tag (e.g. “Corporate Workspace account at [company],” “Personal Gmail account”), so a personal request goes to your personal account and a work request to your work account — never the wrong one.
- It picks the right mailbox automatically. It uses those labels to choose the account that fits each task instead of defaulting to one.
- Calendars stay sorted too. Lookups cover all your connected calendars at once, while still tracking which event lives where.
Sensible boundaries
- You decide what’s connected. Disconnect anything anytime at /integrations.
- It’s intentional, not sloppy: it reasons about which account fits the request rather than mixing them blindly.
- Some scheduled routines can be set to run once per account when you want each mailbox handled separately.
Bottom line: context carries across your accounts, but your Townie stays aware of which account each thing belongs to.
Where does it get its context?
What it draws on. The everyday things you already use Town with — email, calendar, chats, and contacts. It learns about you and gains context based on work you’re already doing. Your writing style and key facts about you are learned automatically based on work you’re already doing.
The more you connect, the more it knows. Every integration you add (Slack, Notion, Drive, Linear, GitHub, and more) becomes another source of context. The more you connect, the more context your Townie gains — and the more helpful it can be. You control what’s connected at /integrations.
Does memory carry over between Slack, email, the app, and the Mac app? It all carries over. Whether you reach your Townie from Slack, email, the web app, or the Mac app, it’s the same assistant with the same memory — so you never have to repeat preferences when you switch.
Do I have to teach it things, or does it pick them up on its own? Mostly it picks things up on its own — your style and key facts come straight from how you already use it. For specific preferences, just tell it once and it remembers.
Managing what your Townie remembers
You can also teach your Townie directly, and it remembers across every routine run and every channel.
Global vs. routine-specific memories
- Global memories apply to everything your Townie does, across all routines — e.g. “I prefer bullet points over paragraphs,” “My work hours are 9 AM to 6 PM Pacific,” “My manager is Sarah (sarah@company.com).”
- Routine-specific memories apply to a single routine; your Townie figures out which one from context — e.g. “Include San Francisco weather in my morning briefings,” or “Keep Friday afternoons free when suggesting meeting times.”
When a routine runs, it automatically receives all global memories plus any specific to that routine — so telling your Townie “don’t draft replies to recruiters” is respected without editing the routine’s instructions.
Adding and managing memories
Just tell your Townie, from any channel — “Remember to always include the weather in my briefings,” or “From now on, keep Friday afternoons free.” It decides whether the memory should be global or routine-specific based on what you said.
To review or clean up what’s stored, go to Settings → Profile to view and delete memories, or simply ask: “What memories do you have stored?” or “Forget the preference about Friday afternoons.”
Be specific. “I like short emails” is useful; “be better at emails” is too vague. Review your memories periodically and remove anything outdated — and reach for a memory instead of editing a routine’s instructions when you just want to nudge its behavior.
Town