TL;DR: Chris Spear has been running Perfect Little Bites, his personal chef business in Frederick, Maryland, for eight years. He also built Chefs Without Restaurants, a community and podcast for culinary professionals working outside the traditional restaurant model, now at 250 episodes and counting. This conversation covers how he's using AI to run both, how he thinks about reviews as infrastructure, and why most chefs are solving the wrong problem when they market their food.
Chris Spear runs Perfect Little Bites, a personal chef business based in Frederick, Maryland. He started it, found some success, and then everyone wanted to know how. Instead of answering the same questions over coffee indefinitely, he built a Facebook group. A Washington City Paper piece ran the same weekend he mentioned it online. Four hundred people followed the page that week. He didn't plan for a launch.
Someone in the community eventually asked why they weren't doing a podcast. Five years later, Chefs Without Restaurants is at 250 episodes. Episode 250 had Jacques Pépin. The co-host is long gone. The show isn't.
Chris came into this conversation having already built custom GPT models for podcast production and financial analysis. He's not describing the tools in theory. He's describing what they produce.
The origin story is a useful one for anyone thinking about community building in their industry. Chris wasn't trying to build a media company. He was trying to stop having the same conversation over and over. The Facebook group solved a real problem for real people, and it scaled because the problem was real, not because the launch was strategic.
The podcast followed the same pattern. Someone in the community suggested it. Chris had no experience in podcasting. He started anyway. Most podcasts die after 10 episodes. Four million podcasts exist and only 300,000 are active. The ones that survive are the ones where the person making it has something to say and doesn't quit when it gets hard. Episode 250 had Jacques Pépin. That's the benchmark.
The manual way to prep for a podcast interview: research the guest, find past interviews, read anything they've written, build a set of questions, communicate the prep to them before recording. It's hours of work per episode.
Chris's current process: feed everything available on the guest into ChatGPT and generate a starting point. For guests he knows well, this is supplementary. For guests he's never met, it changes the quality of the interview.
He built a custom GPT trained on all 250 episode transcripts. It knows his tone. When a new transcript goes in, it generates show notes, title options for audio and YouTube separately, guest bios with hyperlinks, and an episode breakdown. Not perfect every time. But a fraction of the old time.
The discoverability point: creative episode titles no one can find are worse than plain ones people can. ChatGPT understands that "five ways to make your bar more profitable" is searchable. He doesn't have to make that distinction. The model does.
Chris has tracked his personal chef business obsessively. Customer name, party size, rate per head, tip, mileage. Every month, tabbed in Excel, for eight years.
He took screenshots of all of it and fed it to ChatGPT with one instruction: analyze this.
The outputs: year-over-year revenue growth of 18%. A flag that June consistently underperforms. A note that one client had booked five times in five years and might respond well to an incentive. A plan, based on his own numbers, for how to grow from $80,000 to $100,000 in annual revenue, including what that breaks down to per month.
None of this is magic. It's the kind of analysis a business advisor would charge real money for. What changed is that Chris had the data and uploaded it. That's the whole move.
He also asked ChatGPT to help him build the models. Describe what you want it to do. Ask what it needs. Build from those instructions. The computer talks to the computer. It took time to learn. He's still getting better.
Chris reads the reviews of competing personal chefs in the DC area and feeds that text into ChatGPT. Not the star rating. The actual words people used.
For a restaurant operator: find the best-reviewed place in your market. Paste their Google reviews in. Ask what people are actually valuing. The answer is usually not what you'd expect. It's the bartender who remembered someone's name. The host who made them feel seen. The experience, not the menu. That's your roadmap.
You're not guessing. You're reading a dataset that was written for you, for free, by the people who matter most.
Chris leaves every dinner with a small card printed on nice card stock. It has a QR code that goes directly to his Google review page. He also leaves a basket of his signature cookies, individually wrapped, one per guest.
The cookie is not incidental. Jason names it in the episode: reciprocity. When someone gives you something, the pull to return the favor is real. The cookie creates that pull right after the meal, when the experience is fresh. The QR code is there to capture it.
The card message: positive reviews help his business grow. They help clients like you find chefs like me. Brief. No guilt.
Platform choice: Google only. Not Yelp, which suppresses reviews from people without established profiles. Not Facebook, which not everyone has. Google is universal. He also has a live widget on his website pulling Google reviews directly, so visitors see real-time proof, not curated copy-paste.
Chris spent his early years marketing his food. The techniques, the sourcing. The marketing reflected that.
What the reviews actually said: guests weren't talking about the dish. They were talking about how Chris made mom's 80th birthday feel special. The laughing around the island. How accommodating he was. The food has to be good. That's the floor. But it's not the ceiling.
His advice to newer chefs: if you're hitting eight farmers markets to source a vinegar that cuts into your margin and most guests won't notice, that's a choice. The chef who shows up fully present often gets better reviews than the chef who sourced everything obsessively. Eight years of client data behind that.
Q: Do I need to pay for ChatGPT to use it for business analysis?
Chris pays $20 a month. That's what unlocks the custom GPT features. The free version is useful. It's not the same depth.
Q: How do I build a custom GPT with no technical background?
Ask ChatGPT to build it. Describe what you want the model to do. Ask it what it needs. Follow those instructions. Chris's starting point was knowing nothing. The model walks you through it.
Q: What if my competitor has thousands of reviews and I have dozens?
That's the asset, not the problem. You now have a free research report on what the market values. You don't need their volume to learn from their feedback.
Q: Is Yelp worth the effort?
Chris's experience: reviews from people without established Yelp profiles often get suppressed. That's effort lost for the reviewer and proof lost for you. Google is universal. The reviews stick.
Q: What if you sense the guests didn't love it?
Not every dinner gets the cookie and the QR card. Chris reads the room. The system is designed to capture genuine enthusiasm, not manufacture reviews.
Find this episode of the Hospitality Strategy Lab podcast on Apple Podcasts, Spotify, and YouTube.
Chris runs the Chefs Without Restaurants community and podcast for culinary professionals working outside traditional restaurant structures. If that's you, find his work at chefswithoutrestaurants.com.
For independent operators who want to take the AI and review concepts Chris describes and build them into a real system for their venue, The Ops Wire covers this regularly. Subscribe at jlittrell.com. That's where these ideas get translated into something you can actually run.
Jason Littrell spent 10 years behind the bar in NYC, including Death & Co, and served as USBG NYC president. He now runs his hospitality consulting firm entirely on AI. He hosts the Hospitality Strategy Lab podcast and writes The Ops Wire newsletter.
Jason