Products

A protocol-driven knowledge and learning layer. Retrieval combines vector similarity with Personalized PageRank over a typed-edge knowledge graph; a Thompson Sampling bandit on top learns from accept and reject feedback instead of staying frozen.

Install it as an MCP server or call the same surface over REST. One backend protocol runs it anywhere: SQLite on a laptop, Postgres + pgvector or Memgraph in production.

Every phase is instrumented. OpenTelemetry spans and Prometheus metrics feed Grafana, so you can watch the agent learn and catch it when it learns the wrong thing.

Read real text and tap any word for its meaning and native audio; the words you look up become your vocabulary. An AI tutor stays in character and names every mistake by type (grammar, vocabulary, syntax), so you learn what kind of wrong you were.

Real morphology, not pattern matching: CLTK and Stanza for Latin, plus our own extensions to ensk.is for Old Norse normalized toward Icelandic. Point it at any reading and it generates the exercises, dialogs, and vocabulary.

MCP server that gives any coding agent a full LinkedIn publishing pipeline. Voice profiles learn from approve/reject signals and converge on your voice over time.

You approve, it posts, all without leaving your terminal. Your craft is the input; a social presence is the output.

Seven FastAPI services compose into one federated GraphQL supergraph, each a thin, composable skill the gateway exposes. Clients hit a single typed endpoint and never learn how many services answer it.

A new service joins the supergraph and its capabilities appear at the same endpoint, so the client, and any agent driving it, gains new powers without new wiring. Functional alpha.