Back to vendors
P

Potpie AI

Also known as: Potpie, Potpie AI, potpie.ai, potpie-ai/potpie

Visit site
Coding agentindependentVerified 2026-06-30

Open source platform that turns a codebase into an ontology first knowledge graph, then runs prebuilt and custom agents for debugging, testing, code generation, and review in enterprise systems.

Potpie is a platform for building AI agents that understand a codebase deeply enough to change it safely. Rather than dumping files into a model, it parses a repository into a property graph capturing every file, function, class, import, and call relationship, then layers in source history, decisions, tickets, and team knowledge. That ontology first graph gives agents project specific context, so they can answer questions, plan changes, debug failures, and write code that follows the team's conventions. Founded in late 2023 by Aditi Kothari and Dhiren Mathur, the company spent nearly two years building this layer before launching.

On top of the graph sits a suite of agents. Potpie ships prebuilt agents for debugging, test generation, code review, feature building, system design, and onboarding, and lets teams create custom agents through a chat interface. An agent router reads each prompt and dispatches it to the right agent, and agents independently choose and use tools to reach a goal without constant oversight. The approach is spec driven: an agent turns a requirement into an implementation plan, maps dependencies and edge cases, and aligns tests before writing code. Potpie also creates context as systems evolve, updating documentation when pull requests open and drafting designs when tickets are filed.

Potpie connects to the tools teams already use. It indexes and acts on GitHub repositories, pull requests, issues, and commits, syncs with Notion, Linear, and Jira style trackers, assists directly in Slack, and installs its context skills into coding harnesses such as Claude Code, OpenAI Codex, and Cursor. Its architecture is command line first, usable by both people and agents, with a local daemon and a graph explorer. Models are flexible: through a single configuration it routes across OpenAI, Anthropic, Gemini, local models, and OpenRouter, so teams can pick a provider per task or self host for privacy. The project is open source under an Apache license, alongside a managed option.

Potpie targets large, complex systems, codebases from around one million lines to hundreds of millions, where the hard part is reasoning across services, dependencies, and production signals rather than generating a snippet. It positions itself for high risk work like root cause analysis, blast radius detection, and non trivial feature changes, and says it works with Fortune 500 and publicly listed companies in regulated industries including healthcare and insurance technology. In early 2026 it raised a 2.2 million dollar pre seed led by Emergent Ventures. It is younger and less certified than the incumbents, but its knowledge graph foundation is distinctive.

Vendor details

Canonical URL

https://potpie.ai

Category

Coding agent

Subcategory

Ontology first codebase knowledge graph and agent platform

Funding status

Independent. Founded in late 2023 by Aditi Kothari, chief executive, and Dhiren Mathur, Potpie raised a 2.2 million dollar pre seed round in early 2026 led by Emergent Ventures, with participation from All In Capital, DeVC, and Point One Capital. The company reports Fortune 500 and publicly listed customers in regulated industries and more than five thousand stars on its open source project. Total disclosed funding is 2.2 million dollars.

Company status

independent

Use cases & customers

Primary use cases

codebase aware agents for large systemsroot cause analysis and blast radius detectionautomated test generationspec driven feature developmentdeveloper onboarding on complex codebases

Target customers

enterprises with large complex codebases (1M to hundreds of millions of lines)regulated industries including healthcare and insurance technologyengineering teams adopting agentic development safely

Deployment options

Open-source self-hosted (Apache 2.0, CLI and local daemon)Local models via Ollama for data privacyManaged / cloud (account-backed features)Enterprise deployment

Integrations

Indexes and acts on GitHub repositories, pull requests, issues, and commits, syncs with Notion, and indexes Linear and Jira style trackers for teams, issues, and projects. Assists directly in Slack channels, runs in VS Code, and exposes a public API and a command line interface usable by both people and agents. It installs context skills into coding harnesses such as Claude Code, OpenAI Codex, and Cursor, and routes across multiple model providers. Agents act by answering questions, planning changes, writing and refactoring code, generating tests, updating documentation and tickets, and drafting system designs.

In practice

Your codebase exceeds a million lines and changes are risky. Potpie builds an ontology first knowledge graph so agents can trace dependencies, run root cause analysis, and detect the blast radius of a change.

You want to standardize on your own models or keep code private. Potpie routes across OpenAI, Anthropic, Gemini, and local models through one configuration and is open source, so you can self host the whole platform.

A new engineer needs to ramp on an unfamiliar service. They ask Potpie's agents how authentication works or how an order flows, and get accurate answers with references drawn from the codebase graph.

Capability coverage

9.5 / 14 capabilities · 68%

Integrations & Tool CallingIndexes and acts on GitHub repositories, pull requests, issues, and commits, syncs with Notion, indexes Linear and Jira style trackers, assists in Slack, installs context skills into coding harnesses like Claude Code, OpenAI Codex, and Cursor, and lets agents choose and use tools, broad named integrations with real action. Full
Workflow OrchestrationRuns an agent router that dispatches each prompt to the right prebuilt or custom agent, agents that independently choose tools and act without constant oversight, and a spec driven flow that turns a requirement into a plan, maps dependencies, and aligns tests before writing code, plus event driven automation, genuine autonomous multi agent orchestration. Full
Knowledge Grounding & RAGParses a repository into a property graph of every file, function, class, import, and call relationship, then layers in source history, decisions, tickets, and team knowledge, giving agents deep project specific context across systems up to hundreds of millions of lines, a codebase knowledge graph that is the headline of the product. Full
Human Oversight & GuardrailsAims for agents that behave predictably and stay auditable, defines agent behavior through Agent files and spec driven plans reviewed before code is written, and offers collaboration permissions and access patterns, a real oversight and predictability layer rather than a hard runtime guardrail enforcement engine. Partial
Security, Identity & GovernanceRuns open source and self hosted with local models for data privacy, supports collaboration permissions and access patterns, and reports deployments in regulated industries including healthcare and insurance technology at Fortune 500 and publicly listed companies, real data residency and access control, though without a public certification or single sign on matrix. Partial
Observability & AuditabilityProvides a local graph explorer, readiness and diagnostic checks for the daemon, graph, and agent skills, and is designed so agents stay auditable, real visibility into the context graph and agent setup, short of a comprehensive agent execution tracing, audit log, and analytics suite. Partial
Memory & State PersistenceRecords durable project learnings and decisions and continuously updates its context graph as systems evolve, but these are part of the knowledge graph counted under knowledge grounding, and no distinct persistent agent memory or checkpoint and rollback product is documented as first class. Unable to verify
Deployment & Data ResidencyIs open source under an Apache license with a command line first architecture and local daemon, can run entirely self hosted and use local models through Ollama for data privacy, and also offers a managed option, a genuine self host and data residency capability under the customer's control. Full
Prebuilt Agents, Templates & PacksShips a suite of prebuilt agents for debugging, test generation, code review, feature building, system design, and onboarding, and lets teams build custom agents through a chat interface, a real prebuilt library and custom scaffolding, though not a broad browsable marketplace of templates and packs. Partial
Triggers & Channel CoverageAssists in Slack, GitHub, the IDE, a command line interface, and an API, and reacts to development events by updating documentation when pull requests open, drafting system designs when tickets are filed, and producing release notes when releases ship, real trigger and channel coverage centered on the software development workflow. Partial
Model Flexibility & RoutingRoutes across multiple model providers through one configuration using LiteLLM, including OpenAI, Anthropic, Gemini, local Ollama models, and OpenRouter, so teams can pick a provider per task or self host for privacy, genuine multi provider model flexibility and routing. Full
APIs, SDKs & MCP ExtensibilityExposes a public API through a service layer and a command line interface designed to be driven by both people and agents, installs skills into external coding harnesses, and is fully open source for deeper customization, a strong API and extensibility surface, though a formal software development kit and a hosted Model Context Protocol server are not clearly documented. Partial
Testing, Debugging & OptimizationShips dedicated debugging agents that isolate root causes and detect the downstream blast radius of a change, and test agents that generate detailed test plans and code for both happy paths and edge cases including end to end tests, a dedicated testing and debugging engine at the core of the product. Full
Browser & Computer UseRuns a local daemon and a command line first architecture where agents parse repositories, execute background indexing jobs, read and write project context, and drive coding harnesses, real code execution and computer use within the development environment, though not general autonomous browser automation. Partial

Recent platform changes

No recent material changes tracked yet.

Pricing

Free and open source for self hosting under Apache 2.0. A managed cloud version offers a free trial. Paid team and enterprise pricing is not publicly itemized and is arranged with the company.

The open source edition is free to self host, and users supply their own model provider keys, so inference cost is pay as you go to the provider. A managed cloud version offers a free trial. Team and enterprise pricing is arranged with the company and not publicly itemized.

Contact onlyMedium variable costFree tierTrial available

Included quota

Open source (Apache 2.0): the full platform to self host, including the codebase knowledge graph, prebuilt and custom agents, the command line interface, and multi provider model routing, with the user supplying model keys and infrastructure. Managed cloud: an account backed hosted version with a free trial and paid plans. Enterprise: custom deployment and terms for large organizations, arranged with the company. Exact paid figures are not publicly itemized.

What is public

Public: the free open source edition under Apache 2.0, multi provider model routing with the user's own keys, and a managed cloud free trial. Not public: paid managed and enterprise dollar pricing and minimums.

Billing mechanics

Free to self host as open source, with model inference billed by the provider under the user's own keys. A managed cloud tier offers a free trial then paid plans. Enterprise is a custom arrangement. Exact paid figures are not public.

Cost watchouts

When self hosting, the real recurring cost is model inference: Potpie routes to providers like OpenAI, Anthropic, and Gemini using your own keys, so token spend scales with agent activity on large codebases, which can be significant at millions of lines. Running the graph database and daemon also consumes infrastructure. Enterprise deployments are quoted individually.

Variable cost rationale

In the open source edition, Potpie itself is free but agents call external model providers using the customer's own keys, so cost scales directly with usage and codebase size, real variable exposure. The managed tier likely bundles some of this, and enterprise terms are fixed by contract. Overall moderate: the platform price can be zero, but inference spend is genuinely usage based.

Additional watchouts

There is no published paid price, so budgeting the managed or enterprise tiers requires contacting the company. Self hosting shifts cost to model inference and infrastructure, which scales with codebase size and agent activity. The governance surface is less certified than enterprise incumbents.

Overage / add-ons

No Potpie usage meter is documented for the open source edition; model inference is billed by the chosen provider under the user's own keys. Managed and enterprise overage terms are not public.

Sales call required

No — self-serve available

Free / trial

The open source edition is free to self host under an Apache license. The managed cloud version offers a free trial, after which paid plans apply. Exact paid pricing is not publicly itemized.

Lowest paid plan

Not publicly itemized. The lowest cost path is the free open source edition, self hosted with the user's own model keys; the managed cloud version starts with a free trial before undocumented paid plans.

Commercial notes

Potpie's commercial model splits a free open source edition, aimed at developers and teams that will self host and supply their own model keys, from a managed cloud tier and custom enterprise deployments for large organizations. The lever that matters most economically is model inference spend on large codebases, not a Potpie seat price, since the platform routes to external providers.

Key ambiguities

The paid managed and enterprise prices are not published, so the cost above the free open source edition is unknown without contacting the company. For self hosting, the dominant variable is model inference spend, which depends on usage and codebase size rather than a Potpie list price.

Cancellation / refund

The open source edition carries no contract. Managed and enterprise terms are arranged directly with the company and not publicly documented.

Missing data

Managed and enterprise dollar pricing, seat definitions, and any usage minimums are undisclosed. Only the free open source edition and the existence of a free trial are clearly public.

Verified 2026-06-30

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.

Agentic AI Index

A directory and comparison resource for AI agent platforms, autonomous workflow tools, and enterprise agentic automation products.

© 2026 Agentic AI Index

3801 N Capital of Texas Hwy, Ste E240 · Austin, TX 78746

Researched from public vendor sources. See Methodology.