Graphide
Graphide provides a CI/CD service and AI-assisted code review platform that runs GitHub Actions on its own optimized runners, delivering 10–20× faster test times by tracing code changes through a call graph and running only the tests each change can reach. Its review agent reads the full context of a pull request—callers, routes, and data flows—and can verify suspected bugs by running tests against both old and new code. The platform also maintains a versioned graph of the entire codebase, giving developers architectural visibility as AI agents write code.
- Developer Tools
- Artificial Intelligence
- AI Agents
- Software Only
Funding
Raised to date
$500KRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
Founders
Founder details are not available yet.
Product
Problem
AI agents write code faster than anyone can review it, generating thousands of lines of diffs across dozens of files per prompt. Developers face an impossible choice: read every line or trust the agent, while the design in their head and the code on disk silently drift apart with nothing showing where. Traditional CI also runs every test on every change, wasting minutes on tests that cannot be affected by the modification.
Solution
Graphide provides a code intelligence platform that traces a repository into a graph of what calls what, including tests, and uses that graph to power both CI and pull request review. Graphide CI runs GitHub Actions on Graphide's own runners with warm caches, and each change runs only the tests that can reach it—typically 10–20× faster than GitHub-hosted runners. Graphide review gives an agent the diff plus everything the change reaches (callers, routes, tables, services) and your rules, and when it suspects a bug it runs a test on both old and new code, posting the failure in the comment. The graph is versioned with the code, committing, diffing, and merging alongside it, so pull requests carry their architectural context. Every fact the platform reports is marked as guessed, from the code, seen in tests, or seen in production, with a proof that replays.
Target Audience
Primary users are developers and engineering teams shipping production code with AI agents who need to review agent changes, maintain architectural control, and reduce CI wait times without reading every line of diff.
Features
- Multi-language support: Go, TypeScript, JavaScript, Python, Java, Elixir, SQL, Nix, HTML, CSS, and Circom, with every symbol becoming a node with a one-line description
- Agent integration: Claude Code and Codex run inside Graphide, and the graph acts as an MCP server for any MCP-compatible agent
- Versioned graph: the architecture commits, diffs, and merges alongside code in the repository, so pull requests carry their architectural context
- Deterministic compliance checks: static analysis verifies every symbol, HTTP route, and critical edge exists in code, returning the same answer every time with an exit code for CI
- Test reachability tracing: each change runs only the tests that can reach it, with the run page listing what was left out and why
- Self-tuning CI agents: Graphide's agents continuously adjust runners, caches, jobs, and slow tests on the customer's behalf, with nothing to review or merge
- Evidence-backed review: the reviewing agent sees callers, routes, tables, and services the change touches, and can run a live test on old and new code to verify a suspected bug
- Scriptable CLI: `gr init` builds the graph, `gr sync` keeps it current, and `gr check` verifies it for git hooks or CI jobs