Using AI to accelerate hiring and productivity at Zapier
Lessons from Zapier’s experience building hundreds of internal AI agents to reduce engineering friction.
Welcome to the latest issue of Engineering Enablement, a weekly newsletter sharing research and perspectives on developer productivity.
Announcements:
🗓 Laura is hosting a year-in-review roundtable next week with developer productivity researchers from Microsoft, Google, and GitHub, covering this year’s biggest insights and what engineering leaders should expect in 2026. Sign up here.
📚 Abi just announced his book, Frictionless, with Nicole Forsgren, on developer experience. Get a copy here.
Across engineering organizations, AI is often framed as a cost-cutting tool: automate work, reduce headcount, and run leaner teams. But as more companies move beyond pilots, a different pattern is emerging. When AI is applied to the right problems, it doesn’t shrink engineering teams. It increases the return on every engineer you hire.
Zapier is a clear example of this shift. I recently had the chance to interview Andrew Kordampalos, who leads Zapier’s AI Agents transformation. Zapier has built an internal ecosystem of lightweight AI agents that automate thousands of operational and administrative tasks across engineering, product, and internal operations. Their experience points to a simple conclusion: as AI increases the value of each engineer, it becomes more rational to hire more engineers, not fewer.
This article looks at how Zapier reached that conclusion and how they built a culture that treats AI as a force multiplier for hiring and value creation, rather than as a headcount-reduction strategy.
Using AI to remove friction around engineering work
Zapier’s starting point was not “How do we replace developers?” but “Where does work slow developers down?” When Kordampalos joined, his team had grown out of a startup (acquired by Zapier) that originally built real-time meeting transcript agents. Their team quickly turned towards a bigger question: Would it be more effective to automate the choreography around engineering rather than the core engineering itself?
With this line of thinking, they formed a hypothesis that engineers would become far more productive if the friction around them dissolved through agentic automation. The goal was not to change what engineers do, but to change how much of their time is spent on work only they can do.
Zapier’s AI Agents team introduced a network of lightweight internal agents that automate the coordination and overhead surrounding development work:
Agents that collect async standup updates and summarize them, allowing teams to replace five daily standups with two weekly sessions.
Agents that run onboarding steps, including generating email signatures and triggering access to tools, reduce onboarding time to roughly two weeks—significantly less than the industry standard 30–90 days.
Agents that manage Slack workflows, approvals, and operational routines, cutting down on context switching and interruptions.
The underlying mindset is simple: “When you find you’re doing the same thing again and again, you build the automation.” Zapier already had a culture of “build the robot.” AI agents gave that culture a larger surface area.
How AI agents changed Zapier’s hiring math
As Zapier deployed more internal agents, they saw a compounding effect. New hires reached productivity faster. Existing employees increased their throughput without extending their working hours. The organization started to experience AI not as a way to substitute for engineers, but as infrastructure that made every engineer more effective.
This is where the economics shift. If AI removes 10–15% of the administrative and coordination work from an engineer’s week, their effective output goes up. The cost of hiring remains the same, but the value per hire increases. In that scenario, reducing headcount undercuts your own leverage.
Zapier leaned into this logic. As Kordampalos puts it, “We are doubling down and hiring even more people because we want to boost our productivity by using AI.” AI didn’t replace engineers; it replaced the parts of engineering work that prevent engineers from doing their best engineering.
The move from summarizing meetings to a full agent ecosystem
Zapier’s path to a broad agent ecosystem began with a narrower use case. Through the acquisition of Vowel, Kordampalos had already been working with early AI models that transcribed and summarized meetings in real time. That work raised a broader ambition: to make automation conversational, accessible, and embedded in daily work, rather than something only specialists configured. “We realized that the future of creating automations wouldn’t be drag-and-drop interfaces,” he says. “It would be natural language, that’s the universal interface.”
Instead of relying solely on traditional workflows, teams at Zapier began building small agents using their own platform to handle repetitive internal tasks: managing approvals, posting updates, summarizing meetings, and even generating email signatures for new hires. Over time, “There are more bots than humans at Zapier” moved from a joke to a fair description of reality.
For most engineers, the biggest drag on productivity isn’t code, it’s meetings, as I further explore in this article about the true opportunities for productivity improvement with AI. When Kordampalos’s team grew, he noticed they were spending too much time on daily standups. So they built a pair of agents: one that pinged each team member for their async updates, and another that summarized them for everyone. Almost overnight, the daily standup became a twice-a-week sync. “It’s not always one magic automation that replaces your work,” he explained. “It’s the orchestration of smaller agents and bots that you have.”
The insight that multiple small agents can work together like a team (referred to at Zapier as a “pod”) has guided Zapier’s approach. They use Slack as the coordination layer. Agents communicate there, respond to emoji reactions, post updates, and even hand tasks to one another. This ecosystem evolved organically, with employees encouraged to test new agents in sandbox channels before rolling them into production.
On average, Kordampalos says, a new idea takes days, not weeks, to become a working agent.
What Platform and DevEx leaders can learn from Zapier’s model
Zapier’s experience offers a playbook for any platform or DevEx team looking to introduce AI-driven automation inside their organization:
Start with the real bottlenecks. Automate daily stand-ups, status reporting, onboarding steps, and the tasks that steal cognitive energy from engineers.
Keep the feedback loops tight. Give teams space to experiment safely. Launch agents in controlled channels, observe the signal-to-noise ratio, and scale what works.
Build observability and governance early. Create a single dashboard or inventory of agents to manage ownership, access, and performance.
Tie automation to outcomes, not novelty. Focus on measurable gains: reduced meeting hours, faster onboarding, or improved throughput per engineer.
Conclusion: replace or augment?
Zapier effectively tested both paths. They could have treated AI as a way to justify a smaller engineering organization. Instead, by focusing on friction reduction, they turned AI into infrastructure that increased the value of every engineer and made additional hiring more attractive.
Their lesson: AI shouldn’t replace engineers. It should replace the friction around engineers, freeing organizations to hire confidently and get more value from each individual contributor. In a world where AI is reshaping how software gets built, the companies that win will be the ones that use AI to augment their teams and accelerate headcount investment, not the ones that try to do the same work with fewer people.
Who’s hiring right now
This week’s featured DevProd job openings. See more open roles here.
American Express is hiring a Sr. Manager, Digital Product Management - DevProd | Hybrid - London UK
Capital One is hiring a Product Manager - Developer Experience | Plano TX; McLean VA; Richmond VA
Plaid is hiring a Software Engineer - Platform | New York, NY
Reddit: Staff Software Engineer - Developer Experience | Remote - United States
Tesco: Senior Product Manager - Infrastructure | Hybrid - Tesco UK, Welwyn Garden City
Whatnot is hiring a Software Engineer - Platform | San Francisco, Los Angeles, Seattle, NYC
That’s it for this week. Thanks for reading.
 | A guest post by
|
Exceptional framing of the hiring economics around AI productivity gains. The distinction between "replacement" and "friction reduction" captures exactly why most ROI models for AI tooling fail to account for second-order effects. Zapier's approach of treating agents as infrastructure rather than headcount substitutes reveals the compound leverage: when you reduce coordination overhead by 10-15%, you're not just making existing engineers slightly more efficient, you're fundamentaly changing the marginal value of the next hire. What stands out is the pod architecture for agent orchestration. Most organizations deploy AI as isolated point solutions (code completion, meeting transcription) without recognizing that the real productivity unlock comes from inter-agent workflows that automate entire coordination loops. The async standup summarization example is instructive becuase it demonstrates how two simple agents working in sequence can elimnate recurring synchronous overhead. This is different from traditional automation, which typically targets deterministic processes. The broader implication for Platform teams is that developer productivity infrastructure should now include agent orchestration layers alongside CI/CD, observability, and deployment tooling. The companies building this infrastructure today will compound their hiring velocity while competitors are still debating whether AI reduces headcount requirements.