Chris Morrison didn't set out to become the person who runs the systems that run Erewhon. He parked cars. He drove the CEO between stores. He picked up odd jobs wherever the company needed him.
Today, Chris is a Business Analyst and AI lead at Erewhon, the luxury grocery chain with 10 locations across Los Angeles. He runs 89 active Zaps and automations that process roughly a million tasks per year, built a 39-step AI-powered customer service bot that handles 70% of tickets without any human modification, and is teaching company-wide Zapier, automation, and AI adoption seminars.
He doesn't have a computer science degree. He dropped out. And he's one of the clearest examples of what it looks like when one builder, armed with the right tools, transforms an entire company.
The unlikely path to automation
Chris moved to LA to break into the video game industry. He worked as an esports journalist for eight years and parked cars to pay the bills. After a stint as a house car driver at the Four Seasons, he joined Erewhon as a valet.
The store director noticed he was good at more than parking and offered him a new gig: driving the CEO between Erewhon's 10 locations. As they traveled, Chris would pick up work at each store. Cashiering. Valet at another location. Whatever was needed.
Then someone realized he was good at computers.
He started helping the buyers team. Then accounting. Then he built out Erewhon's entire Uber Eats and Postmates delivery infrastructure, connecting all 10 physical stores. Revenue grew, and what had started as side work became his full-time role.
He moved to the Growth and Development team, reconnected with his programming roots (he'd dropped out of college for computer science), and discovered Zapier.
"I would use Zapier as my API back end and webhook back end to connect different systems," Chris said. "That was my introduction to thinking like a programmer."
The moment it clicked
The breakthrough came with a project for Erewhon's buyers team. In the grocery industry, brands apply to get their products on shelves. At Erewhon, those applications were flooding email inboxes. No structure. No consistency. The team was manually converting each email into an Excel line item.
Chris worked with Erewhon's Chief Growth Officer (CGO) to build a series of Zaps that routed supplier data from a Formstack form directly into Airtable. Brands now submit structured information: name, price, Universal Product Code (UPC), brand details. No more manual data entry.
"That was when things really took off for me," Chris said. "Finding the value of automation."
The crown jewel: a 39-step AI customer service bot
Chris had spent a year as Erewhon's own customer service rep. He answered tickets, learned the membership program inside out, and absorbed every policy and process. So he decided to transfer all of that institutional knowledge into an AI-powered system built on Zapier.
Here's how it works:
A customer emails one of Erewhon's 10 store inboxes in Help Scout
The Zap pulls the customer's ID and checks membership status against Erewhon's database
Members get routed to a dedicated Agent with a warmer, more detailed tone; non-members get a standard Agent
ChatGPT reads the message, references a vector store of institutional knowledge, and drafts a response
The system pulls the customer's purchase history from BigQuery and adds it as a note
The draft appears in the store's Help Scout inbox, ready for the manager to review and send
"About 70% of the AI-generated messages are unmodified," Chris said. "The manager looks at it, says yeah, that's good, and fires it away."
Building with AI to prove the AI works
When Chris's boss asked for KPIs, Chris didn't argue. He built a second Agent.
"My boss was like, who's using this? I don't think people are using this. Where are your metrics?" Chris said. "I'm like, I know people are using this. Okay, I'm gonna make it."
Drawing on his background as a writer and journalist, Chris created an "English Tutor" AI agent. It works like a plagiarism checker: it compares the original AI draft to the final response the human agent sent and scores how much was changed.
100 percent similarity: the support manager sent the AI's draft essentially verbatim
70 percent similarity: the support manager kept the structure but modified the recommendations
Below 70 percent: the support agent substantially rewrote the response
He ran the evaluation across hundreds of tickets.
The numbers
AI draft acceptance rate: 70 percent response sent completely unmodified, 100 percent are sent with only slight modifications
Labor hours saved (CS only): 1,500/year
Equivalent cost savings (CS only): ~$40K/year
ROI on CS savings alone: 5.5x return
Steps in the workflow: 39
Store locations automated: 10
"Those are hard numbers," Chris said. "All those MIT studies saying 95% of AI initiatives fail? It's not true. You just have to be creative enough to provide a KPI and provide actual hard facts."
Beyond customer service: 89 Zaps running the company
The customer support workflow is the crown jewel, but Chris has built automation across nearly every department:
Billing and subscriptions: Stripe integrations tracking new customers, cancellations, and updates synced to Google Sheets
Store operations: A 35-step store-to-store transfer system with Twilio SMS notifications
HR and employee services: Mileage tracking with employee verification, phone stipend processing, voided check tracking
Vendor management: Supplier applications, price adjustments, vendor claims, wholesale cost changes through Airtable
Gift card approvals: AI-powered classification of internal requests
Point of sale: Square order processing for kitchen display systems with 24 steps of routing logic
Rapid prototyping: Invoice OCR scanning from concept to working demo in two days and pitched to the CEO
Every single one of the 89 Zaps belongs to Chris. He is the sole builder and saves 5000+ hours or $133k per year, across the business.
"Our company is run on two things," he said. "Vishal, our lead programmer, and Zapier. If one of those two things fail, we're done."
Spreading the automation
Chris is now running a company-wide Zapier seminar. He wants 10 to 20 employees from HR, accounting, and department management to start building their own automations.
His advice: think like a programmer.
"What is my initial data? And then how do I get to the final result?" Chris said. "If this, then this. That's it."
For AI, his guidance is simpler: "Be humble. The fewer assumptions you bring in when you use AI, the more useful it's gonna be for you."
He also has a take most AI evangelists wouldn't expect.
"Controversial take. I think automation is more powerful than AI," Chris said. "AI is probabilistic. Automation is deterministic. Deterministic workflows get you there 99.9% of the time. As it stands today, automation is better than AI."
From parking cars to building the future
Chris Morrison is what it looks like when one person, with curiosity and the right tools, transforms how an entire company operates. He started parking cars at a grocery store. He doesn't have a computer science degree. He taught himself automation, then AI, then software development, building each skill on top of the last.
