Most companies still measure AI by what they roll out: seats bought, licenses assigned, and pilots launched. Those numbers prove that the company has started. They don’t prove anything changed.
The 2026 Zappy Award winners measure it differently. Every one of them moved a number that their business already reports on.
At Galgo, the rate of invalid delivery evidence fell from about eight percent to under two percent across more than 3,000 monthly submissions. At Youtech, $213,000 in phone revenue that had been invisible to ad platforms became attributable to paid search. At Jobber, Ethan Schwandt helped grow the number of active Zapier builders by 122 percent in the first 30 days alone.
And here’s the part worth your attention: these were not company-wide AI initiatives handed down from the top. They started with people close to the work who decided a broken thing didn’t have to stay broken. Every story below is a pattern you can steal.
These are the ten annual winners of the 2026 Zappys.
AI Transformation Leader of the Year: Ethan Schwandt, Jobber
At Jobber, Ethan Schwandt worked with leaders, partners, and builders across the company to make AI adoption practical. He focused on helping employees move from curiosity to useful work through hands-on enablement, job-specific examples, and post-experiment support.
As Senior Manager, Talent Acceleration, Ethan helped connect and scale efforts already underway across Jobber. Working with Senior Leaders and partners across the company, he helped Jobberinos spot where AI and automation could remove a recurring handoff, reduce manual work, or make information easier to act on. Together, that work helped support Jobber’s building culture.
Follow this pattern: Don’t make transformation depend on one expert. Build the conditions for the organization to build.
AI Rising Stars: Carlos Robledo, Hims & Hers, and Doug Hamilton, Klaviyo
Two recruiters. Two halves of the same future. That’s why the Rising Star category has two winners this year.
Carlos got tired of building Friday hiring reports by hand, so he built Rex—a 19-workflow system that connects Ashby, Airtable, and Slack into a live recruiting infrastructure. It now serves recruiters across the organization and can recover up to 50 recruiter-hours a week. Every improvement came directly from the feedback of real users.
The pattern: Build for your own problem first, then put the output where people already work. Adoption follows when the solution is useful before it is official and react quickly to feedback
Doug took a different route. Instead of building the system himself, he built the builders. His enablement program at Klaviyo took the Talent Acquisition team from zero to 57 workflows in three months. 80% of the team adopted Zapier, and ten people were running workflows daily. His AI Tech Stack Tracker surfaced 314 product updates in 90 days, with role-tailored guidance for 61% of them.
Get inspired: Teach people to solve their own problems, then give them a way to keep building. The impact compounds beyond anything one person could ship alone.
Operations AI Builder of the Year: Ignacio Piñeiro, Galgo
Galgo finances motorcycles across Latin America, and a delivery photo is its last chance to catch fraud before money moves. Ignacio built an exception-based review system that pulls delivery data and photos, has GPT-5-mini judge them against strict rules, classifies invalid evidence, and logs every verdict for human audit.
The result: invalid evidence fell from about 8% to under 2% across more than 3,000 photos a month, while human reviewers now focus only on exceptions.
Key takeaway: Point human attention at the risks, not the volume.
Marketing AI Builder of the Year: Justin Hallman, Youtech
Youtech’s clients close real business over the phone—revenue that had been invisible to the ad platforms driving the calls. Justin turned offline conversion tracking into a repeatable agency capability.
The build: Every paid-search click’s GCLID is captured in Zapier Tables. When a quality signal arrives—a qualified lead in the CRM, a completed CallRail event, or a signed client—the system matches it to the original click and uploads the conversion to Google Ads with its revenue value. Delay handling, deduplication, and retries are built in. The core logic stays consistent across clients, even when the ingestion path changes.
The result: $213,000 in previously invisible phone revenue attributed to paid search.
Steal this: Optimize toward the signal that became revenue, not the activity that might become revenue.
Sales AI Builder of the Year: Jocelyne Mendez-Guzman, BioRender
BioRender’s sales team needed follow-up to happen quickly without sacrificing accuracy. Jocelyne built a centralized system where a Gong webhook triggers AI extraction of key points and next steps, while deterministic code assembles exact pricing templates. Slack gives the rep one button, and a Gmail draft appears tailored to the rep’s role.
The result: 3,468 drafts used by 42 reps and CSMs, with follow-up time dropping from 15–20 minutes to about three.
For your next build: Put AI where nuance helps and code where exactness matters. Then measure adoption, not output.
Customer Support AI Builder of the Year: Eric McNulty, Mercari
Mercari’s marketplace generates about 120,000 support tickets a month. Eric built a governed orchestration layer instead of an unsupervised chatbot. New tickets in Dixa are scrubbed, classified against policy definitions, given the right help-center context, and routed to specialized workflows. Low confidence or any failure sends the ticket back to humans.
The result: roughly 47,000 tickets fully resolved by automation every month, 3,000-plus support hours freed, and 97% first-response coverage. The pattern is also available as a Zapier template.
Put Zapier to work where it matters: Build policy guardrails and the human escape hatch first. Then let AI absorb the volume.
IT AI Builder of the Year: Nicholas Francoeur, Compucom
Compucom’s field work lived in tools never designed to talk to each other. Site surveys took a full eight-hour shift, one in four asset entries had errors, and rescheduling a paused site program took 40 hours.
Nicholas built two bridges. AI reads equipment photos and extracts make, model, and serial numbers, which are cross-checked against inventory data. A separate API integration reads spreadsheet data and automatically creates Resource Guru bookings.
The result: surveys down to two to four hours, asset errors from 25% to roughly 1%, and a 40-hour reschedule down to about two.
AI where you need it, deterministic where it matters: Let AI read messy real-world input. Let deterministic integrations carry clean data into systems of record.
HR AI Builder of the Year: Ariel Chen, Figma
A two-person People Ops team at Figma tracks more than 100 background checks a month across 12 countries. Ariel built four workflows that sync Checkr events into an onboarding view and append-only status log. A daily AI triage identifies the specific pending screen, assigns urgency, and routes at-risk cases to senior approvers. Human adjudication and a manual fallback remain part of the design.
The result: monitoring dropped from more than three hours a week to a daily scan of under 15 minutes across 120-plus active candidates.
For your next workflow: In high-stakes workflows, earn trust with testing and logging before you lean on automation.
AI Builder of the Year: Belinda Jimenez, Program Manager
Belinda Jimenez turns recurring project-management bottlenecks into automated systems that help creative teams work independently. Her workflows automate project setup, use AI to make smart resource assignments, and deploy agents to keep projects on track from kickoff to close.
Most importantly, it lets creative pros spend less time waiting on setup and more time doing creative work.
Get inspired: Automate everything predictable. Save AI for the judgment call that resists rules, and keep a human accountability gate.
Start with the winner whose problem looks like yours
If you read one of these and thought, “That’s my problem,” start there. Use Eric’s live support template if support volume is drowning your team. Start with Justin’s GCLID registry if your revenue closes on the phone. Start with Ariel’s tested triage if the workflow you want to automate is one you can't afford to get wrong.
The full story of each winner is live right now.
Congratulations to Ethan, Carlos, Doug, Ignacio, Justin, Jocelyne, Eric, Nicholas, Belinda, and Ariel—the 2026 Zappy Award winners.
