Form follows
intelligence.

Terraphim AI is a privacy-first assistant that runs entirely on your machine. Semantic search across your knowledge repositories, powered by knowledge graphs and Aho-Corasick automata. No cloud. No surveillance. Your data stays yours.

Clone Repository See It Run

Privacy is not a feature. It is the foundation.

Every query, every document, every knowledge graph node stays on your hardware. Terraphim AI processes everything locally using Rust-native performance. No telemetry. No API calls to external servers. Complete data sovereignty.

Knowledge Graphs

Build structured relationships between concepts across your entire knowledge base. Role-specific graphs personalise search to your domain and expertise.

Semantic Search

BM25, TitleScorer, and TerraphimGraph relevance functions work together. Find what you mean, not just what you typed.

Multi-Haystack

Search across local files, Confluence, Jira, Discourse, email, and more from a single unified interface.

Rust Performance

Built from the ground up in Rust with async Tokio runtime. Sub-millisecond autocomplete. Memory-safe concurrency. Compiled to native speed.

Automata Engine

Aho-Corasick automata provide blazing-fast text matching and thesaurus expansion. LeftmostLongest matching ensures precise concept extraction.

Open Source

Fully open source under permissive licence. Audit every line of code. Extend with your own haystacks, relevance functions, and integrations.

One assistant. Every knowledge source.

Terraphim connects to your Logseq notes, Confluence pages, Jira tickets, email via JMAP, Discourse forums, and local filesystem -- all searchable through a single, privacy-respecting interface.

See it in action.

$ cargo install terraphim-agent
Compiling terraphim_automata v0.4.2
Compiling terraphim_rolegraph v0.3.1
Compiling terraphim_service v0.5.0
Installed terraphim-agent v1.5.2

$ terraphim-agent
Terraphim AI v1.5.2 -- Local-first semantic search
Loading role: Terraphim Engineer
Haystacks indexed: 4 (ripgrep, confluence, jira, logseq)
Knowledge graph: 12,847 nodes, 38,291 edges

terraphim> /search "knowledge graph semantic search"
Found 23 results across 3 haystacks (47ms)

1. Knowledge Graph Architecture -- rolegraph design doc
score: 0.94 | source: confluence | matched: 5 concepts
2. Semantic Search Pipeline -- middleware orchestration
score: 0.87 | source: ripgrep | matched: 3 concepts
3. Automata-Based Term Expansion -- thesaurus builder
score: 0.81 | source: logseq | matched: 4 concepts

Built in the open.
Built to last.

Terraphim AI is developed by Applied Knowledge Systems Ltd and released as open source. A Cargo workspace of 29 specialised crates, each with a single responsibility.

From automata engines to agent supervisors, from haystack integrations to knowledge graph orchestration -- every component is inspectable, replaceable, and yours to extend.

View on GitHub Documentation

Architecture for the long term.

Multi-platform: Tauri desktop app, terminal UI with REPL, HTTP API server, and MCP integration for AI development tools. WASM support brings autocomplete to the browser. Firecracker microVMs sandbox untrusted execution with sub-two-second boot.