Zapier, which serves more than three million businesses with automation software, has developed an AI transformation playbook that deliberately sidelines the product. The framework prioritizes organizational structure, accountability and measurable outcomes instead.
Grady Sokaras, Zapier's head of sales, framed the company's philosophy at Zapier Outpost Windsor, a private gathering held at Oakley Court near Windsor from 8 to 10 September. He opened with a stark assertion: "A tool is never the solution." Organizations that purchase software expecting transformation end up with "one more piece of dependency or shelfware."
The retreat brought together roughly twenty enterprise operations and AI leaders. Sessions were led by Philip Lakin, Zapier's director of AI transformation, and Alex Maxwell, who heads Zapier's AI Leaders Lab. Maxwell began with a sobering statistic from MIT NANDA: 95% of enterprise AI initiatives fail to deliver measurable return on investment, despite spending between $30bn and $40bn. The underlying issue, according to Zapier's analysis, is not technological but organizational.
Adoption versus transformation
Maxwell described Zapier's own experience as "performing open heart surgery on ourselves." The company ran hackathons, designated champions, created fluency rubrics and monitored daily usage patterns. Internal adoption reached 100% across all teams, though chief executive Wade Foster has cited 97% in public accounts.
A critical distinction emerged: "Adoption is not transformation. An AI fluent organization that doesn't transform the work itself just gets faster at the same thing." Zapier's argument challenges the prevailing narrative that the problem lies with technology. Instead, the 95% failure rate reflects poor organizational design—specifically, the absence of clear roles, aligned incentives and structured use-case architecture.
The three essential roles
Sokaras outlined a structure that appears consistently in organizations achieving real AI transformation results. Zapier itself operates using this model.
The first role is a C-level mandate-setter who establishes direction and, crucially, imposes constraints. A goal of increased growth without additional headcount works. Vague encouragement to innovate does not. "If there are no constraints and you're just asking people to change, they're not going to change," Sokaras explained.
Second comes an AI transformation guild comprising members from different business units. Guild responsibilities are embedded in their formal job descriptions rather than layered on top, with 30 to 40 percent of their time officially allocated.
Third are empowered practitioners within business lines who identify problems worth solving. They need not build solutions themselves; they must understand what is possible well enough to recognize where processes are broken.
Above these three sits a single AI transformation leader with clear accountability. Sokaras called this position the "single throat to choke."
Three use-case categories and the ambush problem
Role-based use cases create no downstream dependencies. One person accelerates their own work without affecting others.
Team-based use cases differ fundamentally because multiple people depend on the output. Lakin advised bringing operations teams into these initiatives early. Changes upstream can cascade and disrupt downstream work—a phenomenon Sokaras termed "prioritization by ambush."
Organization-level use cases, such as IT helpdesk systems, begin with business systems and operations teams defining requirements. Use cases can migrate upward in complexity. When five separate teams build their own version of the same dashboard, Lakin noted, that signals a systemic need, not five independent solutions.
The $1.2m workflow case study
Zapier's marquee example came from its own operations. Leah Miranda, on the lifecycle marketing team, identified an opportunity: the manual research phase for each sales lead consumed 30 to 45 minutes, and promising prospects grew cold during the delay.
The resulting workflow activates when a target account interacts with company content. It automatically researches the prospect, retrieves contact information from HubSpot and generates a personalized email draft. An account executive reviews the complete package in Slack before any outreach occurs.
The results: $1.2m in pipeline value, 184 deals created, 10 hours per week saved per account executive, 480 annual hours recovered, 4.5 full-time equivalents freed up and $15,000 to $20,000 in avoided tooling costs. These figures come from Zapier and have not been independently verified. Lakin identified this case as the turning point—a high-value, cross-functional initiative visible to leadership that convinced skeptics of AI's potential.
Pushback from the room
The most pointed challenge came from someone with prior experience running similar transformations during the robotic process automation era. That person noted a familiar pattern: a champion receives 20% of their time, but within a month their manager refills the calendar. Lakin's response was direct: the executive alignment was never genuine in that case.
"Unless you create the time and you show that you actually value it, what you compensate, what you incentivize, and what you create time for, it's never gonna happen," Lakin said.
The vendor's offer
Every attendee can engage Zapier's forward-deployed engineering team to build one use case at no cost, regardless of whether it uses Zapier's platform. Sokaras framed the rationale: "Working with AI is a human change management exercise, and people need fast rewards to feel like it is something that is for them."
Source: The Next Web
