Out the Weeds, home

Blog

The Shape of Things to Come: Hospitality Tech When Software Becomes Abundant

AI is making software cheap to build. In hospitality, that won't kill the tech stack. It will change which parts are valuable, how they are bought and sold, and who wins. Here is what the next two to five years could look like for operators and the providers who serve them.

A Tube ad that says a lot

On the Underground this week I saw an ad for an AI app builder. The pitch was simple: describe the app you need, and AI builds it. It wasn't aimed at developers. It was aimed at the lawyers, accountants and ops managers going into the City.

It stayed with me because it captures a real shift. A few years ago a custom app meant a brief, a budget and months of waiting. Now a reasonable first version can take an afternoon. Building software is quickly becoming a commodity.

For an industry that runs on a patchwork of POS, rota, payroll, stock, booking and HR systems, that raises an obvious question: what happens to hospitality tech when anyone can build software?

Building gets cheap. Running well doesn't.

The cost of making a tool is falling towards zero. The cost of making it work in a real business is not.

The hard parts of hospitality tech were never the screens. They are:

  • Clean, trusted data. Sales, labour and stock figures that agree with each other and arrive on time.
  • Integrations. POS talking to stock, rota talking to payroll, payroll talking to tronc.
  • Governance. Who can change what, where staff data lives, and what happens when the person who built the tool leaves.
  • Adoption. Getting tired teams on a Friday night to actually use the thing.

It's telling that the AI app builders sell control as hard as they sell speed. Release sign-off, access rights and security are front and centre. The value is moving from building software to running it well.

The next two years: everyone becomes a builder

By 2028, building small tools will feel as normal as building a spreadsheet does today. Ops managers, finance leads and head chefs will put together their own checklists, trackers and workflows. Many of these tools will be throwaway, and that's fine.

Three changes follow:

  1. AI gets built into the systems operators already use. Most vendors will ship assistants and agents within their products. Having AI stops being a selling point and becomes the minimum.
  2. Buyers ask about outcomes, not features. "Does it have a rota module?" becomes "Will it take two points off my labour percentage?"
  3. Unofficial tools spread. Tools built without IT's knowledge multiply across estates. Some are brilliant, and many create data and compliance risks that nobody is tracking.

The next five years: the agent becomes the interface

By 2030, a GM won't open four systems to understand a bad Tuesday. They'll ask: "Why was GP down at Clapham on Tuesday?" An agent will pull from POS, stock, rota and invoices, and come back with the answer and a suggested fix.

When that happens, the stack splits into two layers:

LayerWhat sits thereWhat happens to it
Systems of recordPOS, payments, payroll, stock, HRDull, regulated, hard to replace. Worth more, because every agent depends on its data.
Everything on topDashboards, reports, simple workflows, point toolsGenerated, customised or absorbed by agents. Cheap to make, hard to charge for.

Pricing follows. Per-site and per-seat models come under pressure when agents do the clicking and fewer people log in. Expect more pricing tied to outcomes, such as labour saved, waste cut or margin recovered.

What it means for operators

  • Large groups build more themselves. Estates of 50+ sites will run smaller, more capable in-house tech teams. They'll build their own front-end tools on top of a few core platforms.
  • Independents want it working from day one. Smaller operators will keep buying packaged software, but they'll expect it to be set up, connected and usable with almost no training.
  • Margin management runs all the time. Agents watching sales, labour and stock will spot GP leakage the same day, not in next week's report.
  • Data quality becomes a board-level issue. An agent is only as good as the data it reads. Operators with fragmented, untrusted data will feel the gap widen.
  • The human parts stay human. Hiring, culture, service, and getting a team to change how it works are not going to be automated.

The biggest risk isn't falling behind on AI. It's ending up with a mess of half-adopted tools, each with its own version of the truth.

What it means for tech providers

  • Single-purpose tools get squeezed. If an operator can build an allergen checklist or SOP app in an afternoon, a vendor charging per site per month for that has a problem.
  • Owning the data matters more. The providers that hold transactions, payroll and stock data, and open it cleanly to agents, become the platforms everything else plugs into.
  • The demo loses its power. When a buyer can mock up an alternative themselves, slick features win fewer deals. Selling becomes about diagnosing the problem, proving ROI and earning trust.
  • Implementation is the product. Setup, integration and adoption become the main differentiators, not a cost to keep down.
  • The go-to-market model needs rethinking. Per-seat pricing, feature-led messaging and long sales cycles built around the demo all need a fresh look.

What could slow it down

None of this is guaranteed, and hospitality rarely moves first. Three things will set the pace:

  • Adoption. Operators are stretched and cautious, and many still run on spreadsheets and WhatsApp groups.
  • Open data. If data stays locked in vendor silos, agents can't join the dots. Open APIs will separate the leaders from the rest.
  • Regulation. Automated decisions about pay, scheduling and hiring will draw scrutiny. That matters most for the people side of the stack.

Every wave of no-code tools has promised that anyone could build apps. Most of the real work still ended up with IT. This wave lowers the barrier much further, but expect it to change the edges of the stack first and the core much later.

Abundant software, scarce judgement

Software is becoming abundant. What becomes scarce is trust, clean data, and people who understand both the operation and the systems behind it. The winners will be the ones who can turn "what's going wrong in this business" into "what should the systems do about it", and then make it stick.

That's the gap Out the Weeds was set up to fill.

← All posts

Next step

Let's talk.

Whether you run venues or build the tech that powers them, we'd love to hear what's on your plate.