Startup Stories - Mixergy
Episode #2299: Zapier is using AI to sell to AI
Date: March 9, 2026
Host: Andrew Warner
Guest: Wade Foster, Founder of Zapier
Episode Overview
This episode explores how Zapier is adapting its business and development strategies for an AI-driven landscape—specifically the emerging trend of “agent marketing,” where software agents, not humans, are the new end users making purchasing decisions. Wade Foster shares hands-on tactics for building “skills” (AI automations), internal productivity hacks, and the organizational philosophy driving Zapier’s adoption of AI throughout the company. The conversation also covers the shifting economics of SaaS, how AI “war councils” are supporting executive decision-making, and practical advice for startups navigating the new AI-first era.
Key Discussion Points & Insights
1. Marketing and Selling to AI Agents Instead of Humans
- Agents as customers:
- “The agent is actually choosing what products to buy on behalf of a human. And so you're no longer advertising to the human. You are trying to get the agent to say, pick me, pick me, pick me.” (Wade, 00:01)
- Agent marketing skills:
- Marketing to agents requires different priorities; agents favor clear, fast, factual documentation delivered in plain text, over visually appealing content geared for humans. (02:18)
- Experiments and analytics:
- Wade emphasizes rapid experimentation and measurement—frequently querying agents, automating evaluations, and adapting API and documentation design for agent compatibility.
- “A lot of this is very much more art than science right now. People are building evals and things like that to kind of track it.” (Wade, 02:18)
2. Internal Build vs. Buy: Is SaaS Dead?
- Internal tools vs. external SaaS:
- Despite being able to build sophisticated tools internally, Zapier continues to purchase specialized SaaS products (e.g., meeting recording software) for stability, support, and ongoing development.
- “The maintenance on this stuff ... at the end of the day, these products are not that expensive and the teams behind them are just putting so much effort in polishing them.” (Wade, 05:15)
- Resource allocation:
- The engineering team’s time is better spent improving core Zapier products rather than reinventing solutions already robustly handled by other SaaS.
- Software spend remains a “single digit percentage” of overall company costs—headcount is a much more significant investment. (06:44)
3. Skill Creation and the AI "War Council"
- Skills as business utilities:
- Zapier employees, even non-engineers, regularly build custom “skills” (modular automations) to solve specific problems.
- Skills are seen as commoditized, similar to how WordPress themes were in the past.
- The War Council skill:
- “If you have a particular problem or a decision you're making ... you can just say, ‘Hey, I want to invoke the war council to like advise me on this.’” (Wade, 00:25; 07:42)
- The War Council spins up multiple AI personas—e.g., “ruthless CFO,” “wartime operator,” “compassionate customer support”—to debate issues and summarize recommendations, helping avoid human bias and groupthink.
- “Each one gives their response to it ... and then there's a master agent, the head of the council, that sort of summarizes it and says, here's what I think you should do. And then it says, if I had $1,000, here's what I would bet ... is the best answer.” (Wade, 07:42)
4. Empowering All Employees with AI Tools
- AI democratization:
- Zapier encourages all employees, regardless of technical background, to become fluent with tools like Cursor, Claude/Opus, and Codex, treating them as daily work companions or “copilots.”
- “We have this value: Don’t be a robot, build the robot … we try and have a vibrant culture of show and tell.” (Wade, 24:21)
- Internal hackathons & workshops:
- Frequent company-wide workshops and hackathons foster peer learning, accelerating the spread of best practices across disciplines—marketing, sales, accounting, HR—not just engineering. (24:21)
5. Using AI for Advanced Context and Workflow Integration
- Second brain / copilot setup:
- Wade demonstrates how he uses Cursor as a “second brain,” integrating company strategy, product roadmaps, meeting notes, etc., enabling rapid, context-aware queries and skill invocation.
- “I use it as like a second brain ... my copilot has company strategy, product strategy …” (Wade, 15:54)
- Switching between AI models:
- Different models are useful for different tasks—Opus is “more enjoyable” for conversation, but Codex5.3 (OpenAI) is “much, much better” at hard technical problems. Wade treats switching models as akin to consulting multiple experts in a room. (16:42)
- AI-augmented data analysis:
- Feeding complex spreadsheets into AI for summary and insight extraction, then checking the AI’s conclusions with human intuition.
- “It's like putting, like a rocket booster on my back ... I can work so much more effectively with this stuff because it does all the things that I would want to do but just can't because I don't spend all my day doing this stuff.” (Wade, 28:52)
6. Integrating with Zapier’s MCP (Multi-Channel Platform)
- Connecting tools and automating workflows:
- Users can enable AI models to access various productivity tools via Zapier MCP, allowing for complex workflow automations like drafting emails, calendar management, etc.
- “You authenticate your client with Zapier MCP ... it's like big window pops up and says, I grant access and then boop, it can go work with it.” (Wade, 23:10)
Notable Quotes & Memorable Moments
-
On the shift to agent-first marketing:
- “You're no longer advertising to the human. You are trying to get the agent to say, pick me, pick me, pick me.”
— Wade Foster (00:01)
- “You're no longer advertising to the human. You are trying to get the agent to say, pick me, pick me, pick me.”
-
On the War Council AI skill:
- “It's just ... I don't know, like, I feel like a kid playing video games again.”
— Wade Foster (09:42)
- “It's just ... I don't know, like, I feel like a kid playing video games again.”
-
On not rebuilding existing SaaS:
- “We would rather deploy that engineering power to making our stuff better versus, you know, improving our internal tools.”
— Wade Foster (05:15)
- “We would rather deploy that engineering power to making our stuff better versus, you know, improving our internal tools.”
-
On the practical value of skills:
- “They're not hard to build at all … there is value in just seeing what other skills people have built. Because, you know, like, the creativity of these users is ... what's so cool is you're like, ‘Oh, wow, I like the way you did that. I want to use that too.’”
— Wade Foster (11:47)
- “They're not hard to build at all … there is value in just seeing what other skills people have built. Because, you know, like, the creativity of these users is ... what's so cool is you're like, ‘Oh, wow, I like the way you did that. I want to use that too.’”
-
On AI as a productivity multiplier:
- “It's like putting a rocket booster on my back ... I can work so much more effectively with this stuff.”
— Wade Foster (28:52)
- “It's like putting a rocket booster on my back ... I can work so much more effectively with this stuff.”
-
On focus and the ‘No’ hat:
- “We know what we got to go do. Stay focused, like, go execute on the stuff that we said was important. And so he gave me a no hat because I have that exact same problem ... I just get too excited about the new thing, and I need to remember to be like, stop. Like, that's a cool idea, but we have cool ideas. Let's go execute on those.”
— Wade Foster (31:01)
- “We know what we got to go do. Stay focused, like, go execute on the stuff that we said was important. And so he gave me a no hat because I have that exact same problem ... I just get too excited about the new thing, and I need to remember to be like, stop. Like, that's a cool idea, but we have cool ideas. Let's go execute on those.”
Timestamps for Key Segments
- [00:01] – What is agent marketing?
- [02:18] – Tactics for optimizing products for AI agents
- [04:53] – The logic behind buying vs. building SaaS products internally
- [07:42] – The War Council skill: origins and usage
- [10:36] – How to recreate a skill from a meeting transcript using AI
- [14:25] – Real example: using War Council for critical decision support
- [15:54] – How Wade sets up “copilot” and context management in Cursor
- [16:42] – Switching between AI models (Opus, Codex, Claude) for optimal results
- [22:12] – Using Zapier MCP to automate emails and other tasks
- [24:21] – Fostering a culture of AI fluency throughout the organization
- [25:35] – Real-world workflow: using AI for sales data analysis
- [28:52] – How AI amplifies productivity for non-engineers
- [31:01] – The story behind the prominent ‘No’ hat
Takeaways and Advice for Startups
- Prepare for agent-first commerce: Document and design APIs for machines rather than humans—prioritize clarity, speed, and factual detail.
- Don’t waste resources rebuilding infrastructure if excellent tools exist; save your creative capacity for competitive differentiation.
- Empower everyone in your team to “build the robot, not be the robot”—democratize AI by offering training, support, and culture around experimentation.
- Leverage modular “skills” and context-rich AI tools as flexible extensions of your brain and team—these are key to future productivity.
- Beware of distractions; focus, prioritize, and execute—sometimes, a simple ‘No’ reminder is essential, even in the age of endless innovation.
This summary captures the in-depth discussion between Andrew Warner and Wade Foster, providing a roadmap to the AI-forward future of automation, productivity, and organizational focus for startups and established businesses alike.
