FathymOwn your intelligence.

The problem

The model is the commodity - everyone has the same frontier models. What's not commoditized is the context you feed them: your projects, your decisions, your standards, your history, your team's accumulated knowledge. Today it's scattered, re-explained every session, lost when people leave, and quietly leaking into vendor models. Fathym is the layer that captures it, keeps it yours, and makes it compound.

  • Context leaks into vendor models you don't own
  • No shared context between your people
  • Working this way can make your people worse at parts of their job

Claude, Copilot, Cursor - all of them forget.

The solution

Two products, one motion: land with your personal context (fAI), then connect your company's real data and turn the work into shared, owned tools (OpenX).

fAI

Your personal context workbench. One command installs it into every AI agent you already run, and your AI starts remembering your actual work - memory that is yours, not the vendor's.

OpenX

Your company's knowledge becomes a living, governed asset your AI draws on - shared context instead of being re-explained by every employee, lost to turnover, or leaked into someone else's model.

fAI and OpenX are two products, told as one continuous motion: land with the individual (fAI - a personal context workbench, free, in every agent) and expand to the org (OpenX - connect your team's real data to AI and turn the work into shared, owned tools). One architecture underneath, so it scales from a single developer to a whole company without re-platforming. (Mosaic, the ambient interface over all of it, is the horizon.)

Context

The knowledge vault gets smarter every session through synthesis.

Capability

The factory produces better capabilities from real usage and feedback.

Scale

Patterns discovered at team level flow back down as better individual defaults.

deno run -A jsr:@fathym/fai/install

Market

$6.27B -> $28.45B by 2030

Context Engineering (Gartner)

Moat

Why can't OpenAI or Microsoft just do this? Their business model is the opposite of letting you own and remove your context - Microsoft folds AI into Copilot, Windows, and Azure; the model vendors keep your context inside their own chat. We're the independent layer that runs between every provider, and your context accumulates in a repo you own - so when the labs ship memory, you become the neutral substrate they all plug into, and the asset compounds where you keep it: switching cost that grows for you, not against you.

Most AI work resets every session. The Loop keeps it: Intake feeds Thoughts, Thoughts feed Workstreams, Workstreams feed Creation, and Creation feeds back in - so the picture of what you know grows instead of resetting.

Traction

fAI live: one-command install across every agentProspect: 6 Word docs -> working dashboardProvider-neutral by designEU AI Act Aug 2026

Team

Matt Smith (CEO) + Michael Gearhardt (CTO)

The context layer that captures your AI context, keeps it yours, and makes it compound. We land with the developer - fAI installs into every agent they run - and expand to the company.

If the labs ship memory: we are the neutral substrate they plug into; context compounds in your repo, not theirsProvider lock-in: multi-model and git-portable by architecture, not by promiseEarly stage: the wedge is live and shipping; this raise funds the path to paying teams

fathym.com

investors@fathym.com