Continue the conversation — chat opens pre-seeded with the current signal, caps, and movement.
Sourcegraph's AI coding assistant, leveraging their decade of code intelligence expertise. Primary differentiator is deep integration with Sourcegraph Code Search, providing multi-repository, codebase-wide context that competitors cannot match. Following the Amp spinoff (December 2025), Cody is now Sourcegraph's sole AI coding product, focused exclusively on enterprise customers after discontinuing Free and Pro tiers in July 2025.
Amp relationship: Amp spun out as independent company Dec 2025 with Quinn Slack as CEO. Amp targets individual developers seeking frontier autonomous coding; Cody serves enterprise teams requiring stable, compliant AI infrastructure. For fully autonomous agentic coding, Sourcegraph directs users to Amp—not Cody.
January 2026 agents: New AI coding agents announced (Code Review in early access; Migration, Testing, Documentation, Notification agents coming). Signals pivot toward greater autonomy while maintaining enterprise focus.
Market position: Mindshare declined from 7.2% to 3.8% YoY (PeerSpot Jan 2026) vs Cursor at 25.7%. No new funding since $125M Series D (July 2021) at $2.625B valuation. Revenue ~$50M (Mar 2025).
Adoption & Proof Points
- Major Customers: 4 of 6 top US banks, 15+ US government agencies, 7 of 10 top public tech companies. Named customers include Leidos, Qualtrics, 1Password, Coinbase, Palo Alto Networks, Booking.com, Nutanix, Lyft, CERN, F5, SoFi, Yelp, FactSet, Nine.
- Leidos Case Study (Fortune 500, National Security): 75% reduction in time spent answering teammate questions (8 hrs/week to 2 hrs/week). 50% reduction in legacy code orientation time. Database migrations in minutes vs full sprints. Air-gapped deployment using self-hosted Llama.
- Palo Alto Networks (AWS Partner case study): 2,000+ developers onboarded in 3 months, 20-40% productivity gains, goal to add 1,500 more users.
- Nine (Australian media): $276K annual savings, 1,200 hours saved.
- Nutanix: 4-day Log4j remediation across entire codebase.
- Workiva: 80% reduction in batch change time.
- Booking.com (DX partnership study): 16% higher PR merge rate for daily Cody users.
- Government Pathway: Exclusive Leidos alliance (Feb 2024) for US government. Leidos $14B+ revenue, serves DoD, FAA, and classified missions.
Recommended Use Cases
- Organizations with complex, multi-repository codebases where cross-repo context is critical
- Enterprises requiring stringent compliance (SOC 2 Type II, ISO 27001) with flexible deployment options
- Teams already using Sourcegraph Code Search who want unified code intelligence
- Government-adjacent organizations working through Leidos pathway
Risks & Limitations
- Mindshare dropped from 7.2% to 3.8% YoY (PeerSpot Jan 2026) while Cursor commands 25.7% and GitHub Copilot holds 7.3%. Competitive pressure intensifying.
- $59/user/month with 25 developer minimum ($17,700+/year). No path for individual developers or small teams—Sourcegraph directs them to Amp. No free tier for evaluation creates procurement friction vs Copilot/Cursor.
- Dec 2025 spinoff shifts agentic focus to Amp (separate company). For fully autonomous task completion, Sourcegraph explicitly directs users to Amp—not Cody. May limit Cody's autonomous capability investment.
- Last funding: $125M Series D (July 2021) at $2.625B valuation—nearly 5 years ago during peak ZIRP. Revenue ~$50M, team ~187 employees. Operationally stable but no recent capital in AI arms race.
- Government deployment requires Leidos partnership and self-hosted configuration. Direct FedRAMP authorization path unclear. Limits certain federal procurement.
- Neovim extension marked "experimental" with no active maintenance expected. Risk for teams with Neovim standardization.
- Agentic features focus on context-gathering, not fully autonomous task completion. New agents (Jan 2026) in early access—not production-validated. For agent-mode, users directed to Amp.
- Higher price point than alternatives ($59 vs Copilot $19), though includes Sourcegraph Code Search value.
Capabilities & Integration
AI Agents (Jan 2026): Code Review Agent (early access), plus planned Migration, Testing, Documentation, and Notification agents with Agent API. Signals strategic pivot toward autonomy.
Context Intelligence: Cody's standout feature is codebase-wide understanding via Sourcegraph's code graph. Retrieves context across entire organizational codebases—100,000+ lines of code awareness through RAG-based retrieval. Supports up to 10 repositories searched simultaneously via @-mention.
Agentic Context Fetching: Autonomously gathers and refines context before responding. Proactive retrieval from codebase, project structure, terminal (Enterprise, with permission), and external services via MCP. Context window ~7,000 tokens for LLM, but RAG system aware of 100K+ lines.
Auto-edit (VS Code, Jan 2026): Context-aware edit suggestions based on recent changes—experimental feature adding autonomous value.
MCP Integration: Anthropic launch partner for Model Context Protocol (Nov 2024). Supports Postgres, Linear, GitHub, Jira, and custom MCP servers. MCP OAuth authentication added Jan 2026. Advantage over Copilot/Cursor/Amazon Q which lack native MCP.
Multi-LLM Flexibility: Claude 3.5/3.7 Sonnet, GPT-4o/GPT-5, Gemini 1.5/2.0. BYOK support for Azure OpenAI, AWS Bedrock, Google Vertex AI—avoiding model lock-in.
Core Features: Chat, autocomplete (single and multi-line), inline edits, custom prompts/commands, code explanations, test generation, documentation generation, Deep Search (including Slack integration).
IDE Support: VS Code (primary, full parity), JetBrains (GA, all major IDEs), Visual Studio (experimental), Neovim (experimental, no longer actively maintained), and Sourcegraph web app.
