a16z · Video summary

Why Enterprise Software Is Being Rebuilt Around AI Agents

AI agents are transforming enterprise software, moving beyond simple chatbots to handle complex, end-to-end tasks in critical areas like procurement. This shift offers startups a unique advantage over incumbents by focusing on comprehensive problem-solving across disparate systems.

a16z · October 2, 2026 · 59-min episode · 1-min read

Why Enterprise Software Is Being Rebuilt Around AI Agents

Key takeaways

💡AI-native startups have an advantage over incumbents by owning the entire end-to-end arc of a process, rather than being limited by existing systems of record.03:01
📊Most enterprise procurement work happens outside of ERP systems, involving numerous stakeholders, emails, and spreadsheets, which AI agents can integrate and manage.05:28
🧠The four types of AI agents (retrieval, process, policy, principal) represent increasing levels of judgment, with incumbents typically lagging in the more complex, judgment-based categories.06:18
🚧Incumbents are held back from deploying more complex AI agents by internal incentive conflicts and the challenge of selling new, transformative products that might cannibalize existing workflows.10:04
🤝Building trust for AI agents, especially in high-stakes tasks like negotiation, involves starting with human-in-the-loop approaches and gradually increasing autonomy as the agent learns and proves its reliability.16:49
💸Small frictions in large-scale procurement projects, such as a single part arriving late, can cause hundreds of millions of dollars in damage, highlighting the critical need for AI-driven coordination.22:24

Chapters

0:10The Opportunity for AI-Native Startups
0:26LEO's Multi-Agent System
0:27Fine-Tuning Models for Specific Use Cases
0:27Why Incumbents Are Held Back
0:41Building Trust in AI Agents
0:41Direct vs. Indirect Procurement
0:58The Four Types of AI Agents
3:21Procurement in the Physical World

The digest

Enterprise software is undergoing a fundamental transformation, driven by the rise of AI agents. No longer confined to basic retrieval or chatbot functions, these agents are increasingly capable of handling complex, end-to-end workflows, offering a significant advantage to AI-native startups over established incumbents. This evolution is particularly evident in critical, multi-stakeholder processes like procurement, where the ability to integrate across systems and apply judgment is paramount.

The Incumbent's Dilemma vs. The Startup's Opportunity

Historically, the battle between incumbents and startups has been framed as distribution versus innovation. However, the advent of AI introduces a third dimension: the incumbent's ability to layer AI models onto existing systems. While this creates formidable competitors, AI-native startups retain a crucial edge: the ability to own the entire end-to-end job, unconstrained by legacy systems of record.

Incumbents are often limited to their own data silos. Resolving a customer issue, for example, might require touching billing, chat history, and contracts—information spread across multiple, disconnected systems. An incumbent's AI might only access its specific system of record, providing an incomplete picture. Startups, however, can build solutions that span these boundaries, integrating data and processes across an entire workflow, from legal briefs to trial, or from initial procurement demand to final delivery and payment.