How Zapier’s Wade Foster Is Building the World’s Most Connected AI-First Company: 5 Key Learnings + 4 Critical Mistakes to Avoid
From the SaaStr AI Podcast with Co-hosts Amelia Lerutte (Chief AI Officer at SaaStr) and Guillaume Cabane (Advisor at HyperGrowth Partners)
When Wade Foster, co-founder and CEO of Zapier, sits down to talk AI strategy, you listen. The man who built a $5 billion company on just over $1 million in total funding has transformed Zapier from a simple integration tool into what he calls “the most connected AI orchestration platform.” With customers running 50 million AI tasks in just 20 days and a mandate that all new hires must be “AI fluent,” Foster has concrete insights every SaaS leader needs to hear.
In a recent conversation with SaaStr’s Chief AI Officer Amelia Lerutte and Giam Kuban (Advisor at Hyperrowth Partners), Foster shared the real story behind Zapier’s AI transformation—including the workflows that are actually working, the customer adoption patterns that surprised him, and the mistakes that could sink your AI strategy before it starts.
Wade’s Top 5 AI Learnings for SaaS Leaders
1. The 90% Rule: Perfect Agentic Workflows Are a Myth (And That’s Actually Good)
“The more agentic these workflows, the more imperfect they are,” Foster explains. “If you read through customer briefs generated by our renewal agent, I as a human will notice things that are incorrect. It’s 90% correct, but you got this wrong, you got that wrong.”
The key insight? Don’t aim for 100% accuracy. Instead, design workflows that get you 90% of the way there and insert humans at the crucial last mile. Zapier’s renewal agent collects usage trends, analyzes Gong transcripts, reviews Zendesk tickets, and generates upsell recommendations—all automatically. But the final decision and customer interaction remains human.
This hybrid approach is where the real ROI lives. As Foster puts it: “The agent is basically doing 90 plus percent of the work and leaves the last mile to the account rep to make it happen.”
2. Mix Deterministic Workflows with Agentic Intelligence—Don’t Choose Sides
The biggest mistake SaaS leaders make? Thinking it’s an either-or decision between traditional automation and AI agents. “The people that are getting the most out of AI automation today know how to mix and match determinism with agentic workflows,” Foster emphasizes.
Use deterministic workflows for tasks that need to work the same way every time—like adding leads from your website to your CRM. Reserve agentic workflows for tasks that require intelligence and adaptation—like generating personalized sales briefs or analyzing customer health scores.
“If you get a lead from your site, you want that to get into your CRM correctly. You don’t want it to just guess,” Foster notes. “However, there are certain use cases that deterministic workflows simply can’t do well enough.”
3. Your Customer Base Determines Your AI Adoption Speed
Zapier’s advantage in rolling out AI to their sales team wasn’t just cultural—it was mathematical. “We have an incredibly small sales team proportionate to the customer base. Our sales folks touch probably less than 1% of our total customer base,” Foster reveals.
This meant AI wasn’t threatening jobs; it was enabling the team to touch accounts they never could have managed manually. For your AI rollout, consider the math: Are you replacing human work or augmenting human capacity? The latter will always have smoother adoption.
Interestingly, Zapier’s existing automation customers transitioned to AI workflows faster than newcomers. “Folks that have been building deterministic workflows all along were quick on the uptick to add AI,” Foster observed. The lesson: automation experience creates AI readiness.
4. Prescriptive Selling Is the New Standard—Buyers Expect AI-Powered Interactions
“There’s no excuse for a sales rep or CSM or support rep to not have context about your account,” Foster states bluntly. “If they’re showing up to a call and just being like, ‘Okay, so tell me what we’re doing’—discovery is dead.”
Zapier’s renewal agent generates what Foster calls “prompts for humans”—detailed briefs that tell account executives exactly what they need to know and recommend for each renewal. The AI surfaces usage trends, support tickets, call recordings, and provides specific upsell or downsell recommendations.
This isn’t just efficiency; it’s competitive advantage. Buyers in 2025 expect this level of preparation. Companies still doing manual discovery will lose to those using AI-powered prescriptive selling.
5. AI Fluency Must Be Defined Function by Function—No One-Size-Fits-All
When Foster announced that all new hires must be AI fluent, it wasn’t just a PR statement. “You should probably do a skills test. You should probably evaluate people on these things,” he explains. But the key is specificity.
“Go function by function in your company and say, ‘These are the use cases for sales, these are the use cases for marketing, these are the use cases for engineering, these are the use cases for HR.’ Just set a baseline for what your company expects.”
Zapier’s approach includes regular hackathons every four to six months and weekly show-and-tells where employees demonstrate AI implementations. This creates both accountability and inspiration across the organization.
The Three Categories Driving 50 Million AI Tasks
Foster revealed that Zapier’s explosive AI adoption falls into three main buckets:
Content Creation & Generation: Taking Gong call transcripts and automatically generating case studies. “It’s such a straightforward use case where you can make case study generation programmatic.”
Customer Engagement: Automatically responding to reviews, generating follow-up emails from website leads, and making phone calls or texts based on system events. “We need to do some sort of correspondence with a customer based on some event that has happened.”
Cross-Team Content Automation: Voice of customer reports that aggregate data from support tickets, sales transcripts, and NPS surveys, then generate themes and insights. Zapier even built a chatbot so product managers can ask, “I’m thinking about launching a feature around X. Is this a problem that anybody has?”
MCPs vs APIs: The Protocol Wars Are Just Beginning
Foster’s take on Model Context Protocols (MCPs) offers crucial insight for technical leaders: “I think we’re going to end up having both. APIs are deterministic and cheap. Works the same way every single time. MCP is a lot more flexible, but you’re burning tokens every time.”
MCPs excel at helping agents talk to agents—enabling Claude to interact with HubSpot, Google Drive, or email with natural language. But APIs remain superior for high-volume, predictable tasks. “It feels like it’s a new protocol that opens up entirely new use cases.”
Wade’s Top 4 AI Implementation Mistakes to Avoid
Mistake #1: Believing the Agent Hype
“If you go read X, if you go read LinkedIn, you would think that everybody is deploying agents everywhere. Mostly that is hype,” Foster warns. The companies actually succeeding are using hybrid workflows, not pure agents.
Mistake #2: Skipping the Function-by-Function Definition
“I find that folks are somehow skipping the step. You get the CEO memo: ‘We’re an AI-first company.’ But that’s the only step they do, and then it’s just chaos inside the organization.” Define AI fluency expectations for each role before hiring.
Mistake #3: Choosing Full Automation Too Early
Foster emphasizes keeping humans in the last mile: “There’s a lot of risk today if you don’t do that last mile work.” The technology will improve, but rushing to full automation before AI reaches enterprise-grade reliability can damage customer relationships.
Mistake #4: Ignoring the Cultural Foundation
Zapier’s advantage came from their “don’t be a robot, build a robot” culture. “Disproportionately inside our company, we probably have people that are just more biased to liking this stuff.” If your culture isn’t automation-friendly, address that before rolling out AI initiatives.
The Bottom Line: AI-First Means Hybrid-Smart
Foster’s most important insight? The future isn’t all about replacing humans with AI—it’s about creating intelligent hybrid workflows that combine the reliability of deterministic automation with the flexibility of agentic intelligence.
“The folks that really understand this concept are the ones that are getting the most effective use out of automation today,” Foster concludes. They know when to use traditional workflows, when to add AI intelligence, and crucially, when to keep humans in the loop.
For B2B leaders looking to become truly AI-first, the playbook is clear: start with hybrid workflows, define fluency by function, focus on prescriptive selling, and always design for the 90% rule. The companies that master this balance won’t just survive the AI transformation—they’ll define it.
Wade Foster will be sharing more AI use cases and product updates at Zap Connect, Zapier’s upcoming virtual event. Learn more about implementing these strategies at your company by following Foster’s learnings from building one of the most successful AI-first platforms in SaaS.
