One North Dev is a research-led app studio. We build a small family of AI-native apps — each one a deliberate bet on a new interaction primitive: ambient inference, agentic delegation, on-device memory, voice-first reasoning.
signal extraction from the feed firehose
An agent that watches every source you care about — Twitter, RSS, newsletters, niche forums — and returns one short, contextual brief in the morning. Not a list of links. A page of synthesized understanding, with citations.
domain-tuned models inside a craft tool
Industry-specific AI that knows your workflow — legal contracts, clinical notes, real-estate diligence, financial filings. Not a chat box you paste into. A tool that understands the structure of your work.
long-term recall, on-device, encrypted
A second brain that lives on your phone. Every photo, voice memo, screenshot, and document becomes part of a private graph you can ask questions of — and that gets smarter the longer you use it. Nothing leaves the device.
presence over productivity
A daily companion that's quieter than ChatGPT — checks in, listens, remembers, and stays out of the way. Voice-first, ambient, opinionated. We're researching what "good company" means as a software primitive.
Each direction begins as a research thread or a paper we've been reading. We prototype the idea before we ever discuss positioning, pricing, or category. If the prototype doesn't earn its place on our own phones, it doesn't ship.
One designer-engineer per app, embedded for the full arc. We don't rotate people off after launch — the team that builds an app maintains it. Apps are commitments, not bets the studio walks away from.
We TestFlight for months, not weeks. Launches are quiet — no countdowns, no leaderboard chasing. We ship when the app is ready for the long haul, not when the marketing calendar wants a story.
Our backers are aligned with multi-year horizons. We don't need to extract revenue in the first six months — which means we can keep apps small, focused, and free of the engagement-loop tricks that come with shorter timelines.
The next generation of apps won't be defined by a feature list. They'll be defined by what they understand — about you, about your work, about the moment. Designing for that requires a different posture from feature-shipping.
On-device first. Cloud only when it has to be, and only end-to-end encrypted. We don't train on user content. We don't sell aggregate behavioral data. We don't run analytics SDKs.
Each app is one bet, not a platform play. The temptation in AI is to build the kitchen-sink agent. We're choosing the opposite — small surfaces, deep capability, no menu of toggleable AI features.
Models will keep changing. Frameworks will keep churning. We're building apps designed to be maintained for years, with abstractions that survive a model swap and a UI that doesn't depend on whichever provider is cheapest this quarter.