Bringing inference back to you.
For the last decade we handed more and more to the cloud. It amplified what we could do, and quietly took some things with it -- our latency, our privacy, our sense of how these systems actually work.
We think the part of intelligence closest to you should stay close -- on a device you can see, in a place you can unplug. It won't always be the fastest option, but it should be the one you trust.
What SidebyStar is building is more than a product. It is a foundation we hope others can build on, so that ambient intelligence can grow naturally in every home, clinic and community.
Every home will have a local intelligence hub, the way every home today has a router.
Being traceable and auditable will become a baseline expectation of AI -- the way HTTPS is today.
More collaboration between devices will happen locally, instead of routing through the cloud.
What we're researching
What we've shipped, what is in the lab, and what we hope becomes a standard by 2030.
Visual models that can anticipate
Not only seeing what is happening, but sensing what might happen -- so that an alert can arrive before an incident.
↘ docs/PRODUCT_OVERVIEW.md
Memory that understands causes
Going beyond event logs. We record why something happened, and how to avoid it next time.
↘ docs/CAUSAL_MEMORY.md
Small, intimate, on-device models
Lightweight models running directly on local hardware in homes and facilities -- a kernel of intelligence that belongs to that space.
↘ docs/ARCHITECTURE.md
Provable compliance, no raw data shared
Let facilities prove to regulators that they are compliant, without ever handing over a single line of personal data.
↘ docs/BUSINESS_PLAN.md
Offline profile bundles
Industry and site profiles ship as offline tarballs with manifests and checksums; imports preview first, then admins apply to the fact base and audit trail. Identities and sensitive detail stay at the source.
↘ docs/FEDERATION.md
Roadmap 2026 - 2030
Each step comes with a milestone you can verify.
Stable local fact base and complete audit trail
On-device visual world models and cross-facility experience sharing
Open home-brain interfaces, inviting a wider developer community
A next-generation protocol for devices from different brands
Helping local AI and zero-knowledge audit move toward industry standards