Graphify turns a software codebase into a persistent knowledge graph that engineers and AI coding assistants query instead of grepping raw files. The open-source CLI parses code locally with 36 tree-sitter grammars, writes a queryable graph, and serves it over an MCP endpoint to assistants such as Claude Code, Cursor, Copilot, and Codex.
The hosted platform adds incremental graph builds, Wiki, pull request review, and agent memory in one workspace. Formal verification acts as a merge gate: a solver looks for an input where old and new code disagree, then reports the change proven, changed with the exact breaking input returned, or unknown within its budget.
Graphify sells into the fast-growing market for AI coding assistants, where the constraint has shifted from generating code to giving assistants accurate, current context about a large existing codebase. Its named comparison set is CodeRabbit, Greptile, and Sourcegraph, and its wedge is owning the persistent graph layer those tools query rather than competing as another assistant.
The buyer is the enterprise engineering organization, which is also where willingness to pay is highest, because self-hosting, air-gapped deployment, SSO, and provable merge gates are procurement requirements rather than features. Open-source distribution plus a formal-verification capability gives it a credible path from individual developer adoption to enterprise contracts.
Graphify's graph is built locally by deterministic parsing rather than by sending code to a model, so it needs no embeddings, keeps source on the machine, and returns file-and-line citations an engineer or agent can verify. The graph is shared: the same persistent structure answers both human queries and assistant tool calls over MCP, instead of each assistant rebuilding context per session.
Its formal verification is the differentiator competitors do not match, because it does not merely flag a risky diff but constructs the distinguishing input and returns it as a test to commit. An Apache 2.0 open core with a large public contributor base gives it distribution and trust that a closed reviewer cannot buy quickly.
Graphify prices in four published tiers, using the open-source CLI as the free entry point and monetizing the hosted and self-hosted layers above it. The CLI is Apache 2.0 at no cost with no account, Free gives one developer unlimited repositories with caps on graph size, push builds, PR reviews, and verification runs per month, and Pro at $10 per month yearly removes those caps.
Teams costs $20 per seat per month yearly with a two-seat minimum, and Enterprise is custom-priced per seat for self-hosted BYOC or air-gapped deployments with SSO/SAML, GitHub Enterprise support, custom terms, and a DPA. The pattern is open-core funnel: free local tooling drives adoption, and governance, verification volume, and deployment control drive paid conversion.

Graphify builds an on-device knowledge graph engine that gives AI coding assistants persistent, queryable memory of a software codebase.