A preliminary tally of monarch butterflies that spend the winter along California’s coast looks near the record low. That’s a blow for the beloved pollinator species, which faces an incredibly high risk of extinction.

Showing Companion Core at MIT is my way of contributing infrastructure to that mission: AI you can own, govern, and trust close to the human experience it’s shaping.

If you’re building come find us on Jan 24 and see what local-first experiential AI feels like.

For builders, it unlocks richer persistent experiences without handing the keys to a centralized platform.

Reality Hack is about empowering the people creating experiential systems to guide where this goes.

Local compute means faster response for embodied interaction, reliability even when networks are imperfect, and privacy that’s enforceable by design because the most intimate data (spaces, voices, behaviors, long-term context) can stay with the user.

For embodied, interactive systems, that’s not just a technical tradeoff it’s a governance decision disguised as architecture.

Companion Core is built around a different default: run the “dreaming machine” locally.

That’s where Companion Core fits.

If the next generation of XR assistants and spatial agents depend on sending everything to the cloud, we quietly accept surveillance incentives, policy shifts, higher latency, and fragile connectivity.

This year’s broader EXPERIENTIAL program reinforces that bringing together creators, researchers, founders, and decision-makers to translate frontier tech into deployable systems with societal impact.

Reality Hack exists to push creator agency: hands-on building, interdisciplinary collaboration, and responsibility for real-world consequences.

In other words, they’re becoming “dreaming machines”: systems that participate in constructing reality.

That raises the real question: who controls the machine that shapes the dream and what incentives govern it?

Reality Hack frames a powerful truth: AI, spatial computing, immersive media, and intelligent systems don’t just produce outputs they increasingly shape experience. They synthesize inputs, model possibility spaces, and generate interactions that influence perception, behavior, and agency.