Fail closed
Permission is a property of the graph, not a filter at the end. On doubt, stay silent.
Every company already has a brain. It's just scattered.
The answers exist: in Slack, docs, tickets, and people's heads. We unify them into one graph that knows who may see what and what was true when, then serve it back through surfaces, apps, and agents.
1
permission-aware graph under everything
30+
surfaces, each a read of that one graph
15+
connectors, one-click or by plain-English description
$0
per-token embedding bill, by design
The map
One graph. Everything else is a view.
The graph
The permission-aware, bi-temporal core. Every fact carries its ACL and its history. Everything else is a read of it.
The surfaces
Ask, Search, Decisions, Meeting Prep, Time Machine, and the rest: thin, cited views over the one graph, not bolted-on features.
The apps & agents
A web command center and a desktop Ask-the-Brain app for people; scoped principals for agents, all trimmed by the same ACLs.
Principles
Cite everything
If it can't be traced to a source, it can't be trusted. Every asserted fact carries its provenance.
Know what was true when
Facts are versioned, never overwritten. The brain answers as of any date.
Zero manual upkeep
Consolidation and decay keep the graph fresh on its own. A brain you have to garden is just a wiki.