Software teams move at unprecedented speed, but the systems they operate inside were never built with autonomous agents in mind. Codebases stretch across millions of lines, decisions are buried in scattered tools, and crucial knowledge often exists only in the minds of senior engineers. Potpie emerged to fix this structural gap.
The company has now raised $2.2 million in pre-seed funding to give engineering teams a unified context layer that makes agents genuinely effective inside complex production environments. The round was led by Emergent Ventures, alongside All In Capital, DeVC, and Point One Capital.
The funds will speed up enterprise deployments, strengthen the engineering team, and advance Potpie’s core context and agent infrastructure.
Solving the problem everyone ignored
Aditi Kothari, CEO & Co-Founder of Potpie AI, told TFN: “Most knowledge workers operate inside one primary tool. Designers live in Figma. Finance teams live in spreadsheets. Sales teams live in CRMs. Developers don’t. To ship a feature end-to-end, a developer uses 30 or more tools. Code lives in GitHub. Logs live in Sentry. Tickets live in Jira. Docs live in Notion or Confluence. Context is fragmented across all of them.”
She added, “Traditionally, companies rely on senior engineers to hold this context in their heads. This does not scale, and it breaks completely when you try to introduce AI agents. An agent cannot be effective without the same context a human engineer has. Potpie solves this by unifying context across the entire engineering stack. We pull in information from all these tools, link it together, and make it usable by agents. The idea is simple: an agent is only as good as the information it can access and the tools it can use. We focus on both.”
How was the idea born?
Potpie was founded by Aditi Kothari and Dhiren Mathur, who began working on the problem in October 2023. The founders spent 2 years building the foundational layer that understands codebases and creates the underlying knowledge graph, before launching Potpie publicly last year in January 2025.
Regarding the origin of this idea, Aditi stated, “The idea of building in this space came up in early 2022, right when the first GPT-3 wave was picking up. At the time, most people were focused on using generative AI for knowledge workers like writers, marketers, and analysts. We had a different question: how do we make generative AI actually useful for developers?”
“LLMs were good at text, but they struggled with code. Code is not linear. It is deeply connected, structured, and spread across large systems. So instead of building another coding assistant, we spent nearly 22 months researching and building infrastructure that allows AI to understand software systems the way engineers do. Finally, a year back, we launched Potpie publicly,” she said.
“We built a system that represents codebases graphically, understands what every function and class does, and connects that with context from tools developers already use, like tickets, logs, docs, and reviews. The business model is interesting because we are not just selling a tool. We are building a foundational context layer that makes AI agents usable inside real, complex engineering organisations,” she concluded.
Turning large codebases into navigable systems
Potpie is designed for enterprises with sprawling, deeply interconnected systems—starting around a million lines of code and scaling far beyond that. Rather than acting as another assistant, the platform builds a graphical representation of entire software ecosystems, inferring behaviour across services and generating structured artifacts that allow agents to operate consistently.
It also creates context as systems evolve. As pull requests open, Potpie updates documentation and tickets. When new work begins, it produces system designs. It writes release notes. Its Agent.md files establish how agents should behave inside a codebase, while a tagged, searchable index across APIs, services, and databases drastically narrows the search space.
Early enterprise rollouts demonstrate their impact. One customer with a 40-million-line codebase cut root-cause analysis from nearly a week to about 30 minutes. Another operating decades-old hardware-integrated systems used Potpie to generate and update tests in the background, compressing multiple sprints of work into a fraction of the time.
Today, Potpie works with Fortune 500 and publicly listed companies in regulated sectors such as healthcare and insurtech. Its open-source projects have already crossed 5,000 GitHub stars, signaling strong community momentum.
The road ahead
Reg the future plans, Aditi stated, “Our goal is to go beyond unifying and curating context. We want to help teams actively create context and make their engineering stacks AI-ready by default. In a world where most tools are the Canva of engineering, we are building to be the Adobe. A platform that can handle non-trivial problems, set standards, and support creation and maintenance within the same system. Long term, we want Potpie to be the foundational layer that engineering teams rely on to build, operate, and evolve complex software systems with AI as a first-class participant.”
Furthermore, she also added, “We currently have a team of 12 people and expect to grow to around 18 by the end of the year.”
Anupam Rastogi, Managing Partner at Emergent Ventures, commented: In large enterprises, the real challenge is not generating code, it is understanding the system deeply enough to change it safely. Potpie’s ontology-first architecture, combined with rigorous context curation and spec-driven development, creates a structured model of the entire engineering ecosystem. This allows AI agents to reason across services, dependencies, tickets, and production signals with the clarity of a senior engineer. That is what makes Potpie uniquely capable of solving complex RCA, impact analysis, and high-risk feature work even in codebases exceeding 50 million lines.”
