LoginSign Up
← Back to all sparks
S

Sourcegraph

AI-ASSISTANTS
Velocity5.0

Code intelligence platform with the Cody AI coding assistant for navigating and editing large codebases.

sourcegraph.com ↗

Sourcegraph stopped selling search to developers and started selling retrieval to their agents.

code-searchcoding-agentscontext-retrievalcode-migrationsagent-provenanceenterprise-security
◆Current state
The product line has reorganized around coding agents as the customer. Code Finder runs its own search loop and hands an agent exact files and line ranges; Deep Search now executes scripts in a sandbox against the search APIs so a migration audit returns a CSV checklist rather than thousands of files of tokens; Agentic Batch Changes entered public beta to scope and ship migrations across hundreds of repositories. The argument threaded through all of it is that an agent searching on its own is slower, costlier, and worse than one handed scoped results.
◆Where it's heading
Two consequences of that repositioning are already showing. First, Sourcegraph is pushing into other vendors' agents rather than waiting for its own surface — reranking Claude Code's file picker on symbols instead of path characters is integration work on someone else's tool. Second, scoped retrieval is being sold as governance: if you can prove which files an agent read before it shipped a change, agent context becomes an audit trail, which is the wedge into regulated engineering teams. The security posts extend the same logic from migrations to vulnerability remediation across ten thousand repositories.
◆Prediction
Expect Agentic Batch Changes to move from public beta toward general availability, and more retrieval integrations into third-party coding agents. The provenance angle looks like the basis of an enterprise tier aimed at regulated buyers.

◆Recent moves

  1. 5d ago

    A smarter way to run code migrations with less LLM context

    Deep Search gains sandboxed script execution against the search APIs, so a migration audit comes back as a CSV checklist rather than a token dump of thousands of files. It is a direct application of the context-efficiency argument driving the whole agent repositioning.

    View source ↗
  2. 13d ago

    Make Claude Code's @ picker search symbols, not filenames

    Reranking Claude Code's @ picker so it matches symbol definitions rather than characters in a file path. Small in scope but notable in direction: Sourcegraph is improving retrieval inside another vendor's agent rather than only its own surface.

    View source ↗
  3. 1mo ago

    How to evaluate Sourcegraph on your own codebase

    An evaluation guide separating retrieval quality from agent task completion and cost. Useful buyer methodology, and a hint that customers were conflating the three, but not a product change.

    View source ↗
  4. 1mo ago

    Compliance-first AI: proving agent provenance for regulated engineering teams

    Positioning content arguing that scoped retrieval turns agent context into an audit trail for regulated teams who must prove which files an agent read. It sketches the compliance wedge without documenting a shipped capability.

    View source ↗
  5. 1mo ago

    Code Finder: fast, efficient code search for coding agents

    ⚡ SPARK

    Code Finder is the clearest statement of the new strategy: a search product whose user is a coding agent, not a developer. It runs its own retrieval loop and returns exact files and line ranges, and the later Deep Search and Claude Code work both extend the same premise.

    View source ↗
  6. 1mo ago

    Three places enterprise security breaks down at codebase scale (and why your current tools don't cover them)

    A security argument that vulnerability remediation fails at scale through missing context, detection gaps, and fixes that crawl across ten thousand repositories. It maps the migration thesis onto security work but reports nothing shipped.

    View source ↗