Will you ever build an iOS app?

Maybe. If iMessage stops being enough for what users need, we'll add surfaces. But an app for its own sake isn't on the roadmap.

Why not also WhatsApp?

WhatsApp is dominant globally, and we'll likely add it over time. The US market skews iPhone/iMessage, which is where we're focused first.

Isn't this fragile? You depend on iMessage.

All products depend on something. iMessage is stable Apple infrastructure with documented business APIs via providers. The dependency is no more fragile than depending on the App Store would be.

Why We Built Brumo on iMessage, Not an App

The app assumption

Consumer software has a default playbook: build an app, ship to the App Store, pay to acquire users, try to get 30-day retention above some threshold, raise money on the growth curve.

We looked at that playbook carefully and decided it wasn't right for a personal AI companion. Not philosophically — practically. The economics don't work and the experience gets worse because of the playbook, not despite it.

The install problem

The most broken step in the consumer-AI funnel is install. Between 'user clicks your ad' and 'user sees your first screen' you lose 40-60% of interested traffic. Then another 20-40% between 'sees first screen' and 'creates an account'.

By the time a user is texting your product, you've lost most of the people who wanted to try it. That's a huge tax on a product that is fundamentally about conversation.

Text-a-number skips it. Zero cliff. You click the button, iMessage opens with the first message typed, you hit send, and you're in. Time to first value measured in seconds.

The retention problem

Even when you clear the install hurdle, consumer AI retention is brutal. Users get excited, chat for a week, and then forget the app exists. The app icon sits on the fifth home screen, the user opens it once a month out of obligation, and eventually deletes it.

This happens because apps compete for app-opening. There's a finite budget of 'opens per day' and dozens of products fighting for it.

iMessage is outside that budget. It's already open. Brumo appearing in iMessage isn't a new open; it's using an open the user already made for other reasons. Retention becomes a function of whether the conversation is good, not whether the user can remember to use the app.

The ambient AI bet

Longer-term, we think the best consumer AI lives in the interfaces people already use — not in new interfaces users have to adopt. That's the 'ambient AI' hypothesis: AI woven into your existing life, not a destination.

iMessage is the highest-frequency inbox most people have. If AI is going to be ambient, this is one of the most important surfaces. Building there early is a bet on where the puck is going.

The trade-offs we accepted

Going iMessage-first isn't free. We gave up:

Visual UI. No custom widgets, no charts, no interactive components. Everything lives in text.
Native-app features. No haptics, no camera access beyond what iMessage natively supports, no background permissions.
App Store marketing and discovery. We don't show up in any app-store search.
Direct push. Our 'notifications' are iMessages — which is usually better, but means we can't send rich notifications with buttons.

For a companion product whose job is conversation and memory, we think those trade-offs are correct. For a product that needs to show you data or let you tap through UI, an app would be the right call.

The infrastructure

For technically-curious readers: Brumo runs on Supabase (Postgres, Auth, Realtime, Edge Functions) for the backend, Sendblue for paid iMessage business lines, and Next.js on Vercel for the web dashboard. Replies are generated via OpenAI's API with our own memory layer wrapped around it.

Getting a paid iMessage line is the one non-obvious piece. You can't register one yourself — you go through a provider like Sendblue. The line is tied to a specific product and has different rate-limits and capabilities than a consumer iMessage.

The web dashboard

We built a minimal web dashboard (brumo.app/dashboard) despite the iMessage-first positioning. Reasons:

Billing. Payments have to live somewhere with a UI.
Transparency. Users should be able to see and delete what Brumo remembers. That's hard to do in iMessage.
Integrations. Future Apple Health, AlarmKit, EventKit hooks need a settings page.

But the dashboard is secondary. Most users text Brumo and never open the dashboard. That's by design.

What we'd tell another founder

If you're building personal AI and the core product is conversation + memory, try iMessage-first. The conversion and retention numbers are meaningfully better than an app path at the same product-maturity level.

If your product needs rich UI, agentic visual confirmations, or heavy data visualization — stick with an app. iMessage-first is not a universal answer. It's a better answer for a specific shape of product, which personal AI companions happen to fit exactly.