Pete Koomen: AI as a foundational layer enhances organizational intelligence, empowering finance teams with internal tools, and LLMs democratizing data access for non-technical users | Y Combinator Startup Podcast

Key Takeaways

• AI integration should go beyond being a mere assistant and serve as a foundational layer to enhance organizational intelligence.
• Framing AI as a tool for collective improvement can significantly boost organizational performance.
• Y Combinator (YC) is developing internal tools to allow finance teams to manage workflows independently of software engineers.
• Large Language Models (LLMs) are being used to enable non-technical users to write SQL queries and ask complex questions.
• Centralized databases facilitate more complex and numerous queries, enhancing data-driven decision-making.
• Many companies still rely on outdated Business Intelligence (BI) tools, highlighting a gap in modern data practices.
• Denormalizing data is essential for optimizing retrieval and understanding in AI systems.
• The current era of AI agents is largely “single-player,” with tools designed for individual use.
• A common context layer, such as a data warehouse, is crucial for efficient AI agent operation within organizations.
• YC’s internal tool registry has grown significantly, now hosting over 350 tools to enhance AI agent functionality.
• The evolution of AI tools at YC has allowed for greater adaptability and efficiency in organizational processes.
• The use of LLMs for SQL queries marks a significant shift in enabling non-technical users to engage with complex data systems.
• Centralized data structures are key to maximizing the potential of AI agents in organizational contexts.
• The gap between outdated BI tools and modern data practices underscores the need for innovation in data management.
• The development of internal tools at YC demonstrates the potential for AI to transform organizational workflows.

Guest intro

Pete Koomen is a General Partner at Y Combinator, where he helps lead the firm’s work on AI and startup infrastructure. He previously co-founded Optimizely, which he led to $100 million in annual recurring revenue, and has recently overseen YC’s internal efforts to build agent infrastructure from the ground up.

Building AI-native organizations

• Using AI as a foundational layer can enhance organizational intelligence. “How do you build superintelligence inside a company part of the key thing is not to just use AI as a copilot this is the the thing where you use it as the building layer for everything” – Pete Koomen
• Framing AI as a way for everyone in an organization to improve collectively is incredibly powerful. “If you frame this as a way for everyone in an organization to get better at what they do using the like collective skill and instinct of the people they work with it’s incredibly powerful” – Pete Koomen
• AI should be integrated into the core of organizational structures rather than just as an add-on.
• The role of AI in organizations is to enhance collective intelligence and performance.
• Organizations can leverage AI to fundamentally enhance their capabilities beyond simple assistance.
• AI integration requires a strategic approach to maximize its transformative potential.
• The use of AI in organizations should focus on enhancing both individual and collective performance.
• Building AI-native organizations involves using AI as a foundational element to drive innovation and efficiency.

Empowering finance teams with internal tools

• YC is building internal tools to empower finance teams to manage their own workflows without relying on software engineers. “The original impetus was why don’t we try to build some tools at YC that we could use to run agents that would give the finance team control over their own software” – Pete Koomen
• The development of internal tools at YC marks a strategic shift in software development practices.
• Empowering finance teams with internal tools allows for greater autonomy and efficiency.
• The ability for finance teams to manage workflows independently reduces reliance on technical teams.
• Internal tools are designed to give finance teams control over their processes and improve efficiency.
• The shift towards empowering finance teams reflects a broader trend in organizational autonomy.
• Enabling finance teams to manage their workflows can lead to more efficient and streamlined operations.
• The development of internal tools is part of a larger strategy to enhance organizational capabilities.

Leveraging LLMs for non-technical users

• The initial focus was on using LLMs for writing SQL queries, which allowed non-technical users to ask real questions effectively. “The first thing actually was not agenta coding it was LLMs for writing SQL queries… it worked so well that nontechnical people… could use these tools to ask real questions” – Pete Koomen
• LLMs enable non-technical users to interact with complex data systems effectively.
• The use of LLMs for SQL queries marks a significant shift in data accessibility for non-technical users.
• Enabling non-technical users to ask complex questions can lead to more informed decision-making.
• LLMs have the potential to democratize access to complex data systems within organizations.
• The application of LLMs in finance and tech enhances the ability of non-technical users to engage with data.
• The use of LLMs represents a foundational shift in how organizations approach data interaction.
• LLMs are a key tool in enabling non-technical users to participate in data-driven decision-making.

Centralized databases and data accessibility

• Having all important data in one centralized database allows for more complex and numerous queries to be asked easily. “…we run on our own software and all of that software sits on one postgres database that has everything that’s important to YC’s world in it…when all of that context is in one place…an agent can go and ask any or answer arbitrary questions about our business…it dramatically increased the number of questions that we would ask and dramatically increased the scale and complexity of the questions that we would dare to ask…” – Pete Koomen
• Centralized databases facilitate data-driven decision-making by enabling complex queries.
• The centralization of data enhances the ability to ask more insightful and complex questions.
• A centralized database structure is key to maximizing the potential of AI agents in organizations.
• Centralized data structures allow for more efficient and effective data interaction.
• The use of centralized databases is a critical component of modern data management practices.
• Centralized databases enhance data accessibility and query complexity within organizations.
• The centralization of data is essential for enhancing the efficiency of AI agents.

The gap in modern data practices

• It’s unfathomable that many companies still rely on outdated BI tools in 2026. “…there are people out there watching this who work in places that still use it the majority of people live in that world still and it’s 2026 which is a little unfathomable actually…” – Pete Koomen
• The reliance on outdated BI tools highlights a significant gap in modern data practices.
• Many organizations still use outdated methods, underscoring the need for innovation in data management.
• The gap between outdated BI tools and modern data practices emphasizes the need for technological advancement.
• The persistence of outdated BI tools reflects a broader challenge in adopting modern data practices.
• The reliance on outdated tools underscores the need for organizations to innovate and modernize.
• The gap in data practices highlights the importance of adopting modern technologies to enhance efficiency.
• The need for innovation in data management is critical to bridging the gap in modern data practices.

Optimizing AI systems with denormalization

• Denormalization of data is essential for optimizing agent retrieval and understanding in AI systems. “…you basically have to take that you’re gonna denormalize it and you’re gonna put it in a format that is optimized for agent retrieval and understanding…” – Pete Koomen
• Denormalization enhances the performance of retrieval systems in AI data management.
• The process of denormalization is critical to optimizing data retrieval and interpretation in AI systems.
• Denormalizing data is a key technical process in enhancing AI system performance.
• The role of denormalization in data processing is crucial for AI system optimization.
• Denormalization is essential for improving the efficiency of AI retrieval systems.
• Optimizing AI systems requires a focus on data denormalization for better retrieval and understanding.
• Denormalization is a critical component of effective AI data management practices.

The single-player era of AI agents

• We are still in the single player era of AI agents, where popular harnesses are designed for individual use. “…it feels like we’re still kind of in the single player era of agents where the harnesses that have gotten really popular right claude code codex py openclaw hermes they’re all designed to be used by a single human…” – Pete Koomen
• The current state of AI agents is largely focused on individual use, indicating a limitation in design.
• The single-player era of AI agents reflects a need for more collaborative and integrated solutions.
• AI agent development is currently limited by a focus on individual rather than collective use.
• The design of AI agents for individual use highlights a gap in collaborative capabilities.
• The single-player focus of AI agents underscores the need for more integrated solutions.
• The current era of AI agents is characterized by a limitation in collaborative design and usage.
• The focus on individual use in AI agent development indicates a need for more collaborative approaches.

The importance of a common context layer

• A common context layer, like a data warehouse, is crucial for enabling AI agents to operate efficiently within organizations. “It just turns out is extremely useful there are many tools for connecting individual agent harnesses to you know other MCP tools other other sources of truth but just like a coding agent inside a monorepo just tends to be much more efficient watching our agents operating on our single database that has everything in one schema tells me that there’s a lot of value at least in getting all of the context into one place” – Pete Koomen
• A centralized data structure enhances the efficiency of AI agents in organizational settings.
• The use of a common context layer is key to maximizing the potential of AI agents in organizations.
• A data warehouse provides a centralized structure for efficient AI agent operation.
• The importance of a common context layer is critical for enhancing AI integration within organizations.
• A centralized data structure is essential for optimizing AI agent performance.
• The role of a common context layer is crucial for enabling efficient AI agent operation.
• A data warehouse is a key component of effective AI integration within organizational contexts.

Growth of YC’s internal tool registry

• The internal tool registry has evolved significantly, now hosting over 350 tools that enhance the functionality of AI agents. “We had like 20 tools at the beginning…every time we kind of come upon some piece of work at YC that we think could be improved with an agent we can just add tools and there’s more than 350 today” – Pete Koomen
• The growth of YC’s internal tool registry demonstrates the adaptability and scalability of AI tools.
• The development of internal tools at YC highlights the potential for AI to transform organizational workflows.
• The expansion of the tool registry reflects a commitment to enhancing AI agent functionality.
• The growth of the tool registry underscores the importance of adaptability in AI tool development.
• The evolution of internal tools at YC showcases the potential for AI to improve organizational efficiency.
• The tool registry’s growth highlights the role of AI in enhancing organizational capabilities.
• The development of internal tools is part of a larger strategy to leverage AI for organizational transformation.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.