Sourcegraph Cody review

What it is

Cody is the AI assistant from Sourcegraph, a company that spent a decade building code search and intelligence for very large organisations. That foundation is the product's whole argument: Cody answers questions using a real graph of your code rather than a snapshot of nearby files.

Where it shines

Scale. In an organisation with hundreds of repositories and a decade of history, the hard question is not write this function, it is where is this behaviour implemented, who calls it, and what breaks if I change it. Cody can answer that across the entire fleet because Sourcegraph already indexes it.

For legacy archaeology this is genuinely valuable. Point it at a service nobody has touched in four years and it can explain the call graph, the data flow and the tests that cover it — work that otherwise consumes an engineer's first month.

Enterprise deployment options, model choice and audit controls complete a package clearly designed for platform engineering teams rather than individuals.

Where it falls short

It is aimed at a buyer most readers are not. Setup assumes a Sourcegraph deployment, pricing at the Enterprise tier assumes a procurement process, and the benefits only materialise once your code graph is indexed. For a team of five with one repository, all of this is overhead.

Day-to-day completion quality is also merely average. Developers who use Cody for its context strength often keep Copilot alongside it for typing assistance, which means paying twice.

Pricing in plain English

Free tier with monthly limits. $9 per user per month for Pro. $59 per user per month for Enterprise with full code graph context and self-hosting.

Who should use it

Platform and developer-experience teams at large engineering organisations with sprawling codebases. Small teams should look elsewhere.