Thesis

Intelligence should stay a commons.

Modern Telepathy exists to advance open-source AI—research, models, and systems that make intelligence public infrastructure rather than a private utility.

Open by default

Models, weights, evals, and tooling should be releasable—so progress compounds in public instead of locking behind a few labs.

Inspectable systems

If intelligence is going to mediate how we work and decide, its stack must be auditable: data, training, interfaces, and failure modes.

Builder agency

The people closest to real problems need the right to fork, fine-tune, and ship. Open-source AI keeps capability in their hands.

Shared sense

Telepathy was always a metaphor for thought moving between minds. Open AI is how we build that as infrastructure—not as a closed product.

What we fund and build

We back and ship work that moves open models from demos to durable systems: training infrastructure, evaluation, safety tooling, and products that keep their intelligence layer remixable.

Closed stacks can move fast. Open stacks can move forever. Our bet is on the latter— and on the builders who make shared intelligence feel like a sense we all get to use.