One graph, every project
Profiles, vocabulary, and content memory held on the server and drawn on by every project, person, and agent — versioned and auditable, and learning from every correction. kapi holds the same graph for one project; the difference is reach, not capability.
The context graph →Real-time collaboration
A web editor and a native desktop app connect to the same server: Visual and Table views with content memory and terminology, while edits and presence propagate live to every client.
Collaboration →Connectors
Content platforms, design tools, code repositories, and a developer's checkout are peer routes into one workspace. Most run server-side, with nothing installed and nothing checked out.
Connectors →Content that stays current
A connector sync, a push, or a developer's command starts a server run: reuse what memory holds, draft the rest, check everything, and park what needs a person into the review queue.
Keeping content caught up →