From Firehouse to Grocery Aisle: How One Firefighter Used AI to Kill the Dreaded Store Loop
You know the feeling. You’re halfway through your grocery list, feeling efficient, checking off items one by one. Then you hit the bottom of the list—and realize the milk, the eggs, and that one specific brand of pasta you needed are all back in aisle three. You’ve already passed it. Now you’re doubling back, weaving through carts, muttering under your breath.
It’s a small, daily annoyance. Most of us shrug and suffer through it. But Joe Poynton, a 44-year-old firefighter with 20 years of service in the UK fire service, decided to do something different. He didn’t complain. He didn’t buy a better shopping list app. He built his own.
This isn’t a story about a Silicon Valley founder with a CS degree from Stanford. It’s about a regular guy who used a rapidly growing practice called vibe coding—using AI chatbots to write code he never could have written himself—to solve a hyper-specific, deeply personal problem. And his story holds a powerful lesson for anyone in B2B SaaS: the best product ideas often come from the most unexpected places, and the tools to build them are now in everyone’s hands.
Let’s break down what happened, how he did it, and what your revenue team can learn from a firefighter who got tired of walking back to the produce section.
The Problem: A $0.00 Inefficiency That Drove Him Crazy
Poynton isn’t a tech bro. He’s a veteran firefighter who handles daily emergencies. When he went grocery shopping on his days off, he wasn’t looking for a revolutionary experience. He just wanted to get in, get out, and stop forgetting things.
But his brain, like many of ours, works in a linear list format. He’d start with “bread” (aisle 1), then “apples” (aisle 4), then “chicken” (aisle 2), then “yogurt” (aisle 5). By the time he reached the end, he realized he’d need to backtrack. Rinse and repeat, every single trip.
This isn’t a $100M SaaS problem. It’s a friction problem—tiny, persistent, and deeply human. And for Poynton, it was the perfect candidate for a custom-built solution.
Key lesson for GTM teams: The most painful problems aren’t always the most expensive. Sometimes the highest-value product you can build is one that eliminates a 2-minute annoyance for a specific user segment, 50 times a week.
The Solution: Grocery Flow
Poynton’s answer was an app he calls Grocery Flow. The idea is simple: you input your shopping list, and the app reorders it based on the actual layout of your specific grocery store. No more aisle-hopping. No more doubling back. It builds an optimized route, turn by turn.
But here’s the twist that makes this story relevant to every B2B founder and revenue leader: Poynton didn’t hire a developer. He didn’t learn Python. He didn’t take a single tech course. He used AI—specifically, two large language models (LLMs), Claude and Gemini—acting as his unpaid, always-available, never-judgmental co-pilot.
He described his approach bluntly: “I just started by saying, ‘I’m brand new to this, treat me like an idiot. I don’t know a single word of code. This is my vision. What’s the steps that I need to take to get there?’”
That’s the vibe coding ethos in a nutshell. You don’t need to know how to code. You need to know what you want to build. And then you need to talk to a chatbot like it’s your most enthusiastic junior developer.
How He Actually Built It (The Playbook)
Let’s get tactical. Poynton’s process wasn’t magic. It was a repeatable workflow that anyone—including your non-technical sales reps, your product marketers, or your customer success managers—could replicate.
Step 1: Start with a cheap subscription
He began with a $25 monthly Gemini Pro subscription. That’s it. No enterprise software. No API credits burning a hole in his pocket. He used Gemini to get the basics of app development explained in plain English.
Step 2: Use two LLMs to check each other
This is the nuanced move. Poynton didn’t rely on one AI. He bounced between Claude and Gemini, using each one to troubleshoot the other’s code. When Claude hit a bug, he’d ask Gemini to find the fix. When Gemini gave a clunky solution, Claude would refine it. He effectively turned two general-purpose AI models into a cross-functional development team.
Step 3: Describe the vision, not the code
He didn’t say “Write a Python script that implements Dijkstra’s algorithm on a graph of grocery aisles.” He said: “I want an app that takes my list and tells me the best order to walk through the store so I don’t have to go backwards.” The LLMs translated that human intent into functional code.
Step 4: Iterate in the open
Poynton wasn’t building in a dark room. He tested the app on actual grocery runs. He tweaked. He broke things. He fixed them. This wasn’t a “big launch” product. It was an MVP that got smarter every time he used it.
What This Means for B2B SaaS (Spoiler: It’s Huge)
You might be thinking: “Cute story about a grocery app. How does this help me hit my Q3 targets?”
Here’s the connection. Vibe coding is not just for consumer apps. It’s a paradigm shift in how internal tools, niche solutions, and even customer-facing features can be built—by people who don’t call themselves developers.
For GTM leaders, this changes three things:
1. Your customer success team can build their own workflows.
Instead of submitting a P0 ticket to engineering for a simple data export or a custom dashboard view, a CS manager can vibe code a small internal tool in an afternoon. It won’t be production-grade, but it doesn’t need to be. It eliminates a friction point today.
2. Your product managers can prototype without engineering cycles.
Want to test a new feature idea? Don’t write a 30-page PRD. Vibe code a clickable demo using AI. Put it in front of 5 customers. If they love it, then bring it to engineering. If they don’t, you just saved weeks of dev time.
3. Your sales reps can build custom demo environments.
Imagine a sales rep who says: “I want a version of our product that shows exactly how a manufacturing client would use our inventory tracking.” Instead of waiting for a solutions engineer, they vibe code a stripped-down replica. It’s not perfect. But it closes deals.
The “Civilian Coder” Trend Is Real
Poynton is part of a growing group of civilian coders—ordinary people using AI to solve daily problems with hyper-specific software. Business Insider recently launched a series called “Vibe Code Your Life” to track this trend. And it’s accelerating.
Why? Because the barrier to entry has collapsed.
In 2019, building a simple app required at least a rudimentary understanding of JavaScript, HTML, CSS, database design, and deployment. Today, you can say to an AI: “Build me a web app that takes a list, reorders it based on store aisles, and shows a map.” And within minutes, you get working code.
This is not a futuristic prediction. It’s happening right now, in a fire station in the UK.
What Grocery Flow Teaches Us About Product-Market Fit
Let’s zoom out. Poynton’s app isn’t going to become a unicorn. It’s not designed for mass adoption. It’s designed for one user with one very specific problem. And that’s precisely why it works.
For B2B founders, this is a critical reminder: the best product-market fit is often hyper-vertical and deeply personal. Don’t try to build for everyone. Build for the person who is so frustrated by a specific workflow that they’re willing to learn to code—except now, they don’t have to.
Grocery Flow exists because traditional solutions (Reminders, generic shopping list apps, mental notes) all failed to address the layout optimization problem. That’s a niche. And in B2B, niches are gold mines.
How to Start Vibe Coding in Your Revenue Organization
You don’t need to be a firefighter with a grocery list. You just need a pain point and a willingness to talk to a chatbot. Here’s your 3-step playbook to get started this week:
Step 1: Identify a “double-back problem”
What’s a task your team does every day that forces them to redo work, re-enter data, or backtrack? That’s your grocery store moment. It might be:
- Manually formatting sales call notes for CRM entry
- Cross-referencing two spreadsheets to calculate commissions
- Generating custom reports for different account tiers
Step 2: Describe it to an LLM in plain English
Don’t try to be technical. Be human. Say: “I waste 20 minutes a day copying data from this spreadsheet to that spreadsheet. I want a tool that does it automatically.” Let the AI ask clarifying questions. You’re the product manager. It’s the builder.
Step 3: Iterate in 30-minute sprints
Vibe coding works best when you treat it like a conversation, not a project. Spend 30 minutes building. Test it immediately. Break it. Ask the AI to fix it. Repeat. By the end of a week, you’ll have a functional tool that solves a real problem—built by someone who never wrote a line of code in their life.
The Bottom Line
Joe Poynton didn’t set out to disrupt the grocery industry. He set out to stop walking back to aisle four. But in the process, he demonstrated a new model for software creation: AI-enabled, human-vision-driven, and radically accessible.
For B2B leaders, the message is clear. You don’t need to hire more engineers to surface every internal inefficiency. You need to empower your existing team—sales, CS, marketing, ops—to become vibe coders. Give them a $25 subscription, a clear pain point, and the permission to build.
Because the next great SaaS product isn’t going to come from a 25-year-old founder in a hoodie. It might come from a 44-year-old firefighter who just wanted to get his shopping done faster.
And he built it by talking to an AI, not by writing code.
Are you ready to vibe code your own workflow fix? Start with one problem. Open a chatbot. Describe it like you’re talking to a friend who knows how to build things. See what happens.
Your grocery store—or your sales pipeline—might never be the same.