Potpie AI is a codebase intelligence platform founded in 2023 by Aditi Kothari and Dhiren Mathur, publicly launched in January 2025, and headquartered in San Francisco. It converts source code, tickets, logs, documentation, and code reviews into a unified Neo4j-based knowledge graph, capturing every file, function, class, and the dependency relationships between them. AI agents then reason across this graph rather than operating on isolated file snippets. The platform offers a suite of specialized agents for debugging (automated stacktrace analysis with codebase-specific guidance), Q&A (explaining functions and features in context), low-level design, test planning, and code generation for new features or refactoring. A spec-driven development workflow lets developers describe what they want to build in plain English, and Potpie generates production-ready code that respects the existing architecture and patterns. Potpie is available as a self-hostable open-source project (the main repo has accumulated over 5,000 GitHub stars) and supports multiple LLM providers including OpenAI, Anthropic, Ollama, and OpenRouter. It integrates into developer workflows via a VSCode extension, Slack integration, and native connectors to Notion, Jira, and Calendly. An observability feature called Potpie Trace provides a full audit trail of every agent thought, tool call, and state transition. The company raised a $2.2 million pre-seed round in early 2026, led by Emergent Ventures with participation from All In Capital, DeVC, and Point One Capital. The funding was aimed at expanding the platform's ability to unify context across large engineering stacks. A customer with a 40-million-line codebase reported reducing root cause analysis time from nearly a week to around 30 minutes after adopting Potpie. Pricing ranges from a free individual tier to a $20/month Pro plan, with an early-stage startup promotional tier at $1/month and custom enterprise pricing that includes centralized governance, compliance controls, and optional self-hosting. Key features: - Neo4j-based knowledge graph covering every file, function, class, and relationship in a repository - Spec-driven development: describe desired features in plain English, agents generate architecture-aware code - Specialized agents for debugging, Q&A, test planning, low-level design, and code generation - Self-hostable open-source backend with support for OpenAI, Anthropic, Ollama, and OpenRouter - VSCode extension and Slack integration for in-workflow access without context switching - Potpie Trace: full observability into every agent thought, tool call, and state transition - Native integrations with Notion, Jira, and Calendly for enterprise toolchain connectivity - Custom agent builder for creating repeatable, precision workflows on any codebase task
Free for individuals; $20/month Pro; $1/month early-stage startup promotional tier; custom enterprise pricing