--- title: "Two solo chefs are running restaurants with AI doing their back-office. Here's the stack." description: "Business Insider profiled two solo chefs in late April who are running cooking businesses on AI back-office tooling. The pattern is real and it generalises beyond restaurants. Here's the actual stack a one-person UK restaurant or food business can run on £80 to £400 a month: bookings AI, supplier admin AI, payroll AI, inventory AI, and social-content AI. Named tools (Nory, MillieAI, OpenTable AI, plus the cheaper alternatives), what each costs, the pattern across the two case studies, and the three traps that catch solo founders trying to copy the setup." canonical: https://richardbatt.com/blog/solo-chefs-ai-back-office-restaurant-stack date: 2026-05-05 author: Richard Batt tags: [AI Implementation, Solo Founders, Restaurants, Automation Stack] type: blog_post --- # Two solo chefs are running restaurants with AI doing their back-office. Here's the stack. _Business Insider profiled two solo chefs in late April who are running cooking businesses on AI back-office tooling. The pattern is real and it generalises beyond restaurants. Here's the actual stack a one-person UK restaurant or food business can run on £80 to £400 a month: bookings AI, supplier admin AI, payroll AI, inventory AI, and social-content AI. Named tools (Nory, MillieAI, OpenTable AI, plus the cheaper alternatives), what each costs, the pattern across the two case studies, and the three traps that catch solo founders trying to copy the setup._ **Richard Batt** — AI implementation specialist. 120+ projects across 15+ industries, serving SMBs (5-200 employees) worldwide from Middlesbrough, UK (working globally). Contact: richard@richardbatt.com · https://richardbatt.com Business Insider published a piece on April 29th profiling two solo chefs running cooking businesses where AI handles most of the back-office work. One runs a small private-dining and meal-prep operation; the other runs a single-location neighbourhood place with a team of three on shift. Both told the reporter that AI tooling does the bookings, the supplier admin, the payroll, the inventory tracking, and most of the social content. Both said they couldn't run the business without it at the volume they're running. After 120+ AI implementations across 15+ industries, including five hospitality and food businesses in the last 18 months, the pattern in the BI piece is the one I see in UK restaurants and small food businesses too. So this post is the stack version. What each layer costs, which tools the UK market is actually using, and the three places solo founders waste money trying to copy this kind of setup. **The short version** - The five-layer back-office stack a solo chef can realistically run on AI: bookings, supplier admin, payroll, inventory, social content. - UK monthly cost runs £80 at the cheap end (mostly off-the-shelf SaaS with AI features) to £400 at the rich end (Nory, MillieAI, premium booking AI, paid social-content tooling). - Nory and MillieAI are the two UK-rooted hospitality operations tools that come up most often in this market in 2026. - The biggest waste pattern is buying every layer at once. The chefs in the BI piece both built the stack one layer at a time over six to nine months. - The biggest non-obvious win is supplier admin. The supplier-invoice and order-confirmation workflow is where most solo operators lose three to five hours a week, and AI handles 80% of it cleanly. ## What the BI piece actually documents The two chefs profiled aren't outliers. They're early-pattern operators in a market that's adopting AI back-office tooling faster than most other small business categories. UK hospitality has been squeezed by labour costs, energy prices and rent, so the appetite for tools that take admin off the founder's calendar is real and immediate. The BI piece is short on the specific tools, longer on the working day. The pattern that comes through is consistent. Both founders kept cooking and customer-facing work on themselves and pushed everything that wasn't either of those onto AI. The hours saved sat in the 10 to 20 a week range for each of them. And both said the saving paid for itself inside 60 days at their volume. That last point is what makes this market interesting for SMB AI in general. The break-even is short because the alternative isn't a tool, it's the founder's own time, and the founder's time is the constraint on the whole business. Save five hours a week, the founder serves more covers or sleeps more nights. Both turn into revenue or retention. ## The five-layer stack Below is the layer-by-layer view, with named UK-relevant tools and rough monthly costs at the volume a solo chef or single-location operator would run. The cost bands assume one location, one founder, two to four part-time staff, and 200 to 600 covers a week. ### Layer 1: bookings This is the layer where AI was helpful first and is now table-stakes. Most of the meaningful UK booking platforms (OpenTable and Resy at the top end, SevenRooms for guest profiling, ResDiary for the cheaper end) have shipped AI features in the last 18 months. The features fall into three buckets: predictive availability, automated confirmation and reminder messaging, and last-minute table-fill outreach. The OpenTable AI features are the most widely adopted in this market because most independents are already on OpenTable. The version most solo operators run is the standard OpenTable subscription (about £150 to £230 a month at the volume above) with the AI features included. ResDiary is the cheaper alternative at £60 to £100 a month with a thinner AI feature set. SevenRooms is the more sophisticated end at £200 to £400 a month, only worth it if the founder is doing personalised guest profiling and repeat-guest workflows. The honest read is that AI bookings save the founder about 90 minutes a week on no-show chasing and last-minute reseating. The integration to a phone-line voice agent for off-hours bookings is the next layer most operators add, but I'd hold that until layers 2 and 3 are working. ### Layer 2: supplier admin This is the layer most founders underrate and where the highest hourly saving sits. The supplier-admin job covers ordering, order confirmation, invoice matching, dispute handling, and the goods-in receiving check. Done by hand, it's three to five hours a week for a single-site operator. Done with AI tooling layered onto a hospitality operations platform, it's 30 to 60 minutes. The two tools that come up most often in UK kitchens in 2026 are Nory and MillieAI. Nory is a fuller hospitality operations platform that includes labour scheduling, sales forecasting and supplier admin in one product. The pricing runs £150 to £350 a month for a single site, depending on modules. The supplier-admin features alone are worth more than the cost for most operators above 300 covers a week. MillieAI is the leaner alternative aimed specifically at one-to-three site independents. The pricing is £80 to £180 a month and the focus is the supplier invoice and order-confirmation workflow rather than the full ops platform. If a founder only buys one AI tool in their first six months, MillieAI is a strong candidate because the workflow saving is concrete and shows up on Sunday night, the worst time to be doing supplier admin manually. The cheaper alternative is to wire a generic AI workflow tool (Make.com or Zapier with the AI steps enabled, plus a customised prompt setup against ChatGPT or Claude) into the existing accounting tool. That route runs £30 to £60 a month plus 8 to 12 hours of one-off setup. It works well if the founder is technical or has a friend who is. It breaks if the supplier list is bigger than 10 or if the invoices come in mixed formats. ### Layer 3: payroll Payroll AI is the layer where the technology is good and the regulatory tolerance is low. UK payroll has to be compliant with HMRC and RTI, with auto-enrolment correct on pension. So the AI layer here is almost always wrapped inside an existing payroll provider rather than a standalone product. The three providers I see most often in UK hospitality at this size are Sage Payroll (with its 2025 AI features), Xero Payroll (with the JAX AI integration), and Rotaready (which sits between scheduling and payroll and is used heavily in independent hospitality). Each runs £20 to £80 a month for a single-site operator. The AI value is mostly in the timesheet-to-pay-run reconciliation, which used to take a founder 40 to 60 minutes a fortnight and now takes 10 to 15. If the founder is doing tronc, the AI doesn't help much yet. Tronc allocation rules differ per venue and the AI tooling hasn't generalised them well. Plan for tronc to stay a manual job into 2027. ### Layer 4: inventory and stock Inventory is the layer where AI is selectively useful and frequently oversold. The marketing pitch is that AI predicts your weekend stock from a four-week sales history, places the order automatically, and reduces waste by 20 to 30%. The reality is messier. Where AI inventory tooling actually helps is the routine reorder cycle on stable lines (dry goods, drinks, packaging). Nory does this well as part of the broader operations platform. So does MarketMan if the founder is on QuickBooks or Xero. The cost runs £60 to £150 a month for a single-site operator, often bundled into the platform layer. Where it doesn't help yet is fresh-produce ordering against a daily-changing menu. The model can predict broad volume but not which 12 of 30 mains will be that night's special, and getting that wrong is the difference between a 4% food cost variance and a 12% one. Most solo chefs in the BI pattern still keep fresh-produce ordering on themselves, with the AI handling the dry-goods reorder cycle in the background. ### Layer 5: social and content This is the layer with the most options and the highest waste rate. Every social tool now claims AI features. Most are unhelpful for a small food business because the AI generates plausible content that doesn't sound like the chef and gets ignored by the algorithm. The pattern that works across the operators I've seen is narrower. The chef writes the captions, takes the photos, and uses an AI tool for two specific things: scheduling the post across Instagram, TikTok and Facebook (Buffer or Later, both £15 to £40 a month with AI scheduling), and turning long-form content (a Sunday batch of food photography) into a fortnight of varied posts. The second job is what AI handles well. ChatGPT or Claude with a customised prompt does most of it. The cost is £20 a month plus 90 minutes on a Sunday afternoon. What doesn't work is full-auto AI social content. The two chefs in the BI piece both said the same thing: their followers know when the chef wrote the caption and when a tool did. The drop-off in engagement on AI-written content is large enough that the saved time is wasted because the post does nothing. ## The £80 stack and the £400 stack Here are the two most common UK stacks I see in single-site operations. The £80 a month stack: ResDiary or basic OpenTable for bookings (£70), a Make.com workflow for supplier-invoice OCR plus matching to Xero (£20 plus £25 for Xero), Xero Payroll for the pay run with JAX (£15 add-on), a manual inventory spreadsheet, and Buffer for social scheduling plus a £20 ChatGPT subscription. The founder still does most of the supplier and inventory work; AI shaves the worst hour off each of them. The £400 a month stack: OpenTable with the full guest CRM features (£200), Nory for ops, supplier admin and inventory (£250), Rotaready for scheduling and payroll (£60), and Later or Buffer for social plus a Claude subscription for content support (£40). The founder does almost no admin work in a typical week, only review and exception handling. The break-even is harder at this volume but real if the founder is the constraint on revenue. The middle of the market sits at £150 to £220, usually with MillieAI or Nory's lighter tier handling the supplier and inventory layers and a standard booking platform handling the front of house. ## The three traps that waste money Three patterns waste solo-founder money in this market, and I've seen each one inside the last six months. The first trap is buying every layer at once. The two chefs in the BI piece took six to nine months to build their stacks. The founders I've seen lose money in this market are the ones who tried to deploy the full stack in six weeks because a vendor sold them a bundle. The integration cost on five tools at once is real, the team can't absorb the change at that pace, and at least two of the layers usually turn out to be wrong for the venue. The second trap is buying the platform tier when the lean tier would have done. Nory and the bigger ops platforms make sense at four to six covers per square metre and 400+ covers a week. Below that, MillieAI or a Make.com workflow does the same job for a third of the price. Vendors don't volunteer the tier-down conversation. Ask for it explicitly. The third trap is automating the front of house before the back of house is steady. Every founder I work with is tempted to start with bookings AI because it's visible. Bookings AI saves about 90 minutes a week. The supplier and inventory layers save four to six hours. Start where the time is, not where the demo looks shiny. ## Frequently asked questions ### Is the BI pattern realistic for a small UK restaurant? Yes, with caveats. The two chefs in the BI piece are both solo or near-solo operators with strong personal brands and modest covers per week. The stack works at that scale. It works less well at the 800-cover-a-week independent with a head chef, sous chef and front-of-house manager, because the workflows are more interdependent and the founder is no longer the bottleneck. For a single-site, single-founder UK operation doing 200 to 600 covers a week, the stack is realistic and the cost-saving timeline (60 days to break-even) tracks with what I've seen. ### Which layer should a solo chef start with? Supplier admin if the supplier list is over 10 names and the invoice volume is over 30 a month. Bookings if the venue is below that supplier scale and the no-show rate is the visible problem. Don't start with social or inventory. Both are layers that reward a founder who has time to tune them, and the founder doesn't have time yet by definition. ### How long until AI does the cooking? It doesn't. Or rather, AI does the back-office work that frees up the founder's time to cook, talk to guests, and run the floor. The two chefs in the BI piece were explicit about this. The cooking is what they wanted to keep. The admin is what they wanted gone. AI is good at the second category. It's not good at the first, and the operators who succeed in this market are the ones who keep the boundary clear. ## What I'd do this Monday morning If you're running a UK restaurant or food business and looking at this stack for the first time, the move this week is to pick one layer and build it. The supplier-admin layer is the highest-yield first step for most operators. Try MillieAI on a 14-day trial against your three biggest suppliers. Track the time saved over the first fortnight. If it's three hours or more across the two weeks, the tool will pay for itself before the next quarter ends. The AI Ops Vault has the founder-stack templates, including the supplier-admin prompt setup that runs against MillieAI or against a Make.com workflow if you want the cheaper route. https://richardbatt.co.uk/vault If you'd rather start with the 10-minute version, download the AI Quick-Wins Checklist. It includes the one-page audit I run with hospitality operators to pick the first layer. https://richardbatt.co.uk/quick-wins The two chefs in the BI piece aren't outliers; they're early-pattern operators in a category that's about to look very different inside two years. The stack they're running is the one most single-site UK food businesses will be running by 2027. The founders who start now have a year of compound learning on the ones who wait, and the learning curve here is the kind you build by using the tools, not by reading about them. --- ## More about Richard Batt Richard Batt is an AI implementation specialist who helps businesses deploy working AI automation in days, not months. 120+ projects across 15+ industries. ### Key pages - [Home](https://richardbatt.com/) - [About Richard](https://richardbatt.com/about) - [Blog](https://richardbatt.com/blog) - [Contact](https://richardbatt.com/contact) - [Subscribe](https://richardbatt.com/subscribe) ### Contact - Email: richard@richardbatt.com - Location: Middlesbrough, UK (working globally) - Website: https://richardbatt.com