Zapier Is Using AI to Sell to AI

Key Takeaways

• AI agents now decide software purchases for humans
• Zapier creates plain‑text agent‑focused documentation
• Internal “War Council” skill enables multi‑agent decision analysis
• Zapier prefers third‑party AI tools over building core features
• AI skill marketplace could become new SaaS revenue model

Summary

Zapier founder Wade Foster explains that AI agents are becoming the primary buyers of software, forcing companies to market to machines instead of humans. He describes Zapier’s approach of serving agents with ultra‑fast, plain‑text documentation and internal AI “skills” like the War Council that orchestrate sub‑agents for decision‑making. Foster also notes Zapier’s preference for leveraging third‑party AI tools rather than building core infrastructure, reserving internal development for niche utilities. The conversation highlights a nascent AI‑skill marketplace that could reshape SaaS revenue models.

Pulse Analysis

The rise of "agent marketing" marks a fundamental shift in B2B outreach. As large language models like Claude and GPT act as procurement proxies, traditional human‑centric advertising loses relevance. Companies must expose clean, machine‑readable APIs, minimalistic HTML, and concise metadata so agents can evaluate options quickly. This trend forces marketers to think like developers, optimizing for factual clarity over visual flair, and to monitor emerging evaluation frameworks that track which agents recommend their products.

Zapier illustrates how an AI‑first culture can accelerate this transition. By deploying internal "skills"—lightweight markdown‑driven agents—Zapier enables employees to summon the War Council, a multi‑persona decision engine that debates and ranks outcomes. The company pairs these skills with Cursor, a model‑agnostic IDE, and its MCP integration layer, allowing seamless actions such as drafting emails or updating calendars. Rather than reinventing core functionalities, Zapier leans on third‑party AI services, conserving engineering bandwidth for its automation platform.

The broader implication is the emergence of an AI‑skill marketplace, where creators sell reusable decision‑making modules much like SaaS subscriptions. As agents increasingly mediate software purchases, the value shifts from the underlying application to the quality of the skill that guides the agent. SaaS vendors that expose well‑structured, agent‑friendly endpoints and invest in high‑impact skills stand to capture a new revenue stream, while those clinging to legacy marketing tactics risk obsolescence.