Building intelligence
that compounds.

Primer is an intelligence-development company. Our mission is superintelligence. We believe continual development may be one of the mechanisms required to get there, and Primer is our experimental platform for studying that path.

The starting point is deliberately concrete: teach machines to keep learning after training. That means acquiring new knowledge from experience, retaining it, and building on it. We call this research program the Intelligence Bootloader.

We work like a laboratory, not a product company. Capabilities are claimed only after they are measured, protocols are frozen before experiments run, and negative results are kept alongside positive ones. Primer has not demonstrated AGI or superintelligence, and we say so plainly.

Today’s most capable AI systems stop improving the moment training ends. Everything they will ever know has to be present before deployment.

An intelligence that keeps developing changes what AI can be used for in the near term: tools that adapt to a new domain, retain institutional knowledge, and grow more useful with time instead of going stale.

And in the long term, we believe the ability to accumulate knowledge, connect discoveries, and improve the machinery of learning itself may be what separates very capable models from something greater.