r90.ai

Not a suite of apps.
A stack.

The platform that comes with the team. Context, workflow, and governed agents on one substrate, so the value compounds instead of fragmenting across tools.

Start a conversation See how it fits together

The architecture is one stack of three layers that compound. Canon, the context layer, sits at the base: the company's shared memory, made of artifacts like PRDs, ADRs, epics, research, and decisions. Habitat, the workflow layer, is built on top of Canon and carries the work, ticket by ticket, through spec, build, verify, track, and ship. Olympus, the governed agent plane, acts inside that workflow with a roster of named agents, reading Canon context it never had to ask for, bound by governance. Connections thread between the layers because they share one substrate rather than splintering into separate apps with private memory, so the value compounds with use.

01

Canon, the context layer

Ask

One shared memory, every answer sourced

Every PRD, ADR, and decision the company has made, in one shared memory the whole team can query in plain language. Every answer comes back with its source attached, never a guess, and because it is the base of the stack, everything above it inherits the same memory.

Explore Canon

02

Habitat, the workflow layer

The whole job, in one place.

One living surface where a piece of work carries its spec, its code, and its proof at once, from discovery all the way to production.

Spec, work, and proof on one thread. Five tools collapse into one surface.

Build

One surface, spec to ship

Most tools fragment the work: the spec in one place, the tickets in another, the evidence somewhere else, and the thread between them is lost. Habitat keeps discovery, specs, build, verification, and release on one living surface, so every work item carries its whole thread from spec to ship.

Explore Habitat

03

Olympus, the agent plane

Oversee

Governed agents, with a review council

At the summit, governed agent teams do real work inside the habitat workflow, reading the same canon the team does, never a private copy that drifts. Before anything merges, a multi-model review council checks the work, so no single agent and no single model gets the last word unchecked.

Explore Olympus

The bench

Specialists that plug into the stack.

Beyond the three layers, a growing roster of specialist tools, deployed as each engagement needs them. They read from the same substrate and feed the same workflow.

  • agent usage telemetry

    Tach

    How much AI your team actually uses, and what it costs. Tokens, sessions, and spend across every harness in one view.

  • in-app feedback capture

    Peril

    Point at any element and leave feedback in context. The screenshot, navigation trail, and session replay attach themselves, then it lands in Habitat.

  • accessibility compliance

    Ally

    WCAG checks and vision agents on every PR, with a per-requirement report. Configurable to block the merge or just note it.

Where this goes

The next move is a conversation.

Context, workflow, and governed agents on one substrate. See what the stack looks like against your own canon.

It ships with the engagement, tuned to your stack. The source is yours to keep. No per-seat metering.

Start a conversation (opens in a new tab) Request a demo (opens in a new tab)

30 minutes with a principal, not an SDR.