Launching today
Town
The assistant that learns how you work, then gets to work.
101 followers
The assistant that learns how you work, then gets to work.
101 followers
Town is the unusually helpful AI assistant. It learns how you work, not the other way around: your voice, your people, your patterns. It works where you do, across email, calendar, docs, and messages. It drafts your replies, handles your scheduling, and builds routines for the work you repeat. The more it learns, the more it does on its own, so your time goes to the work only you can do.o big ideas.
Town
I'm Tony, co-founder and CPO of Town. This is the third(!) time I've taken a run at the same idea: an AI that works alongside you. The first was Navigator, which we launched here on Product Hunt back in 2019. The second was at Google, where my team built an AI teammate we demoed on stage at I/O '23. Each time we got closer, but the tech or the timing wasn't there. It is now!
For years the AI race has been about who's most powerful. But for most people AI is still a privilege: it rewards the ones willing to learn prompting, workflows, connectors, MCPs, API keys. Everyone else gets a chat box and a "good luck." We built Town to go the other way around: it learns from the work you've already done and starts doing the work the way you would.
A few things people have actually done with their Townie:
- A CFO drafted 120 individualized investor emails in an afternoon.
- A CPA and working mom cut her busy-season hours from 80 to 60 — Town handled client research and emails, and her kids' school calendars.
- A nonprofit director processes grant requests that arrive as handwritten notes in foreign languages, photographed on a phone. Town translates, transcribes, summarizes, and files them before his first coffee. "It's like having another employee and a half."
- A founder who observes Shabbat mentioned his Friday offline pattern once, in passing. Town built him a Saturday-night briefing without being asked.
No two Townies are alike, because no two people are. It works for you, just you — private, with approval-required defaults, per-action controls, and a full audit log of everything it does on your behalf.
Jean-Denis Greze (my co-founder) and I will both be in the comments all day. Bring your hard questions and your wishlist!
@tonydevincenzi the shabbat story is the one that stuck with me. he mentioned it once in passing and it just built around that. thats the part thats hardest to design for, picking up on something the person didnt explicitly ask you to act on and getting it right without overstepping. genuine question: how does town decide when to act on something it picked up vs just filing it away? that line between proactive and presumptuous feels like the hardest design call in this space.
Town
@dhanishta_likhar Agree. Town (and the Townie) does a surprisingly great job here OOTB, but there's room for improvement. IMO what makes Town work better than any other assistant I've used is that has so much context about you, your work, the people you work with, that the selection of what to remember (and what to forget) is grounded in the scope of what the service does for you. If you're explicit (remember that i don't like blue cheese!) I will remember, but if you are implicit (i don't love afternoon meetings on fri), it will keep a memory bc one of your Townie's jobs is to help with schedule, and it knows that.
The Shabbat story is a perfect illustration of the core design challenge here - picking up on something the user mentioned once, without being asked, and acting at the right moment. That kind of implicit preference learning is genuinely hard to build without it feeling creepy or presumptuous.
I've been thinking about this pattern in meal planning too - I built DishRoll (dishroll.netlify.app) as an AI-powered weekly meal planner that learns dietary preferences and routines over time. The tension is the same: generic suggestions are useless, but the learning phase requires enough signal to be useful. How long does it typically take for Town to move from "helpful assistant" to "actually anticipates what I need"?
Town
@samir_asadov good q: it's immediate (honestly) - during sign up you'll see the beginning of a series of proactive suggestions of work it can take on
Superhuman already handles basic inbox triage for me, so Auto-inbox is the part I’d compare first. Can Town map to my existing Gmail labels, or does it start with its own default set?
Town
@novamaker01 Hey Felix, yes, Superhuman does label triage. Auto-inbox is a bit more advanced, it will research the sender, understand the context of the email thoroughly, then apply labels and/or archive. I am a super heavy "archive everything that isn't a real person and isn't important" user. So my inbox is actually just... signal. To answer you question, you can set the labels for auto-inbox to whatever you'd like. We don't pre-map to gmail, but it's simple to do: just ask your Townie.
Sounds very cool. I'd be super-interested to trial this and see how it fits in with my life - is there free or freemium version to get a feel for if this is a good fit before getting too invested?
Town
@christian_sawyer1 Yes there's a 14 day free trial at sign up (no card or anything required)
@christian_sawyer1 @tonydevincenzi Got it - thanks - gonna check it out
Just found out about this from NYC Tech week during Cristina Ciaravalli's Masterclass on GTM. I signed up yesterday and have been extremely blown away by Town. I'm not technical (no-coding experience, but heavy experience with no-code AI), and I have been AI Assistant-maxxing for the last few years with ChatGPT, Claude, Grok, Openclaw, & one of my personal favorite is Poke by Interaction. I always thought that Poke nailed the proactive part of what I wanted out of an AI assistant, such as speaking to me more like a friend and less like a chatbot, while at the same time writing email drafts in my voice with my context across different apps/integrations. Although I enjoy being able to iMessage an AI assistant like Poke, once you start having ongoing threads and scheduled actions, a single iMessage thread can get incredibly overwhelming to keep track of. From what I've seen so far, I think that Town absolutely nailed the surfaces that need to be seen, like Powers, Library, People, Tasks & Threads. I'm very new to trying it out, but it already feels a lot easier to maintain active threads and ongoing tasks/actions in parallel and I'm excited to continue to work with it. Having the ability to use town through web, iOS, or Mac app was also a MASSIVE plus (although the Mac app was kinda buggy for document editing).
Two questions:
1. Separate Town accounts for work vs. personal, or one with both connected? I signed up with both emails and got the 90-day trial. I'll mostly use Town for work, but if I invest in personalizing my townie, I'd want that relationship to carry forward beyond any single employer. Curious whether you'd recommend a personal account with work connections and work email as secondary added, or keeping them distinct so work context doesn't get crossed with personal.
2. Any plans for Sales/Marketing/BD roles? I saw the open Ops and Engineering roles in SF, plus the note inviting people to reach out if they think they can make Town better. I want to put my hand up for that. Within 24 hours of signing up I'd already organically converted 8 friends to sign up and give it a shot, and I've already received feedback from 2 of them this morning (see screenshots below). Even though I've had Town for less than 24 hours I think I have a general idea of the value-add here and can see this being a game-changer. I'd love to learn more about how you're thinking about go-to-market and the business model, and where someone like me, (not traditionally technical, but skilled using AI and natural language) could plug in.
Ramp
congrats on the launch!!! I love my townie
QuickSheet
Congratz on the launch. Will definitely give this a try