The big interview: Lolly

In the past 18 months, artificial intelligence (AI) has gone mainstream and we’ve all been experimenting with it. Whether you want to simply research something from what is potentially the world’s greatest and largely free encyclopaedia or create a sinister deepfake impersonation for your own nefarious needs, it is hard not to end up both enchanted and petrified by its potential for both good and evil.

But what about hospitality? How does the industry view its potential to transform its businesses? Well, there’s a psychological issue here too, as identified by Peter Moore, chief executive of hospitality technology specialist Lolly.

Operators are fixated on AI, but to some extent it remains elusive – a buzzword that people struggle to get their heads around. AI is not yet being deployed in a grounded, problem-solving way. “We’re looking at it the wrong way,” says Moore. “You need to start with the basics, the problem you want to solve as a business, rather than just talking about AI.”

There is also the fear factor: inaccurate data – or ‘corrupted data’ – is believed to be as high as 40% (at the time of writing), which might also go some way to explaining why many operators are still struggling to monetise AI. As Moore readily acknowledges, large language models (LLMs) – advanced AI systems trained on vast quantities of text – do not necessarily get everything correct. “Today’s LLMs are like talking to the most intelligent person on earth, capable of pulling data from millions of different parts of the world, but they’re still a jack of all trades and master of none,” he says. “You cannot assume everything they say is correct.” Unfortunately, they can recycle bad or corrupted data.

It is therefore easy to see why operators might remain uncomfortable about relying on AI, which might fail, thus risking both reputation and money. But if they wait for zero-risk AI, they might never effectively monetise its potential. And that data is on the verge of becoming a whole lot more reliable. We are moving steadily towards agentic AI, which means that most of the platforms will quickly move away towards specialisation.

“This means that LLMs will individually specialise in a specific subject matter,” explains Moore. “For example, if you want, you can give an AI Gemini, GitHub or Claude AI a task, and they could individually give work to other LLMs to give you the answer or arrange something. Those LLMs will be specialists in what they do, and Claude could act as a project manager to pull everything together.”

On a practical level, you might want to use Claude to work out what sort of holiday might suit you and your family. It will work out that this might be a trip to Disneyland. Claude will then find the right specialist LLMs to book you a taxi, hotel and tickets.

As a result of the move towards agentic AI, data will become far more reliable, asserts Moore. “In contrast to where we are currently, the single agent world, with its growing specialisations, will deliver an immense improvement in the quality and accuracy of data being delivered.”

So, how exactly will all of this impact with operators’ pain points in contract catering? Because AI will surely have to shift the dial on the industry’s extensive challenges if it is to become more than just a talking point. Moreover, it does not take AI to work out what those pains might be: they include workforce management, rising labour costs, squeezed profit margins, demands by younger consumers for digital ordering, waste, allergens and inefficient production methods.

How can AI or agentic AI help resolve these challenges? The fact is that it is already helping with core issues, such as forecasting, data analysis, staff rotas and seasonal menu planning linked to weather. “Our system now does a rolling seven-day forecast, taking in current demand, time of year and weather,” Moore elaborates “So again, this is important because it reduces waste as you're not throwing so much stuff out.”

For contract caterers, that means linking their electronic point of sale, kiosk  and app sales histories to a forecasting model that pulls in day-of-week, seasonality and weather, as well as the ability to automatically generate production and order guides. Put all this together with a more reliable agentic platform and you lift everything up a notch further. A master AI controller calls out to specialist services – perhaps one forecasting demand, one pricing ingredients, one optimising menu mix – then you combine the results to end up with a fiendishly clever plan for both chef and buyer.

Much of this will rest, however, on the ability to combine AI with existing technology, though as Moore freely acknowledges, the solutions to such day-to-day challenges do not necessarily involve “burning the planet”. However, sometimes it’s the integration of both that delivers the best outcomes. “We can identify from the data the flow of staff coming in, so that you can basically manage your workforce in a better way by placing the right people at the right time in the right locations.”

As Moore readily concedes: “You might well be able to achieve all of that with your existing technology. You don’t necessarily need to have AI, but if it can enhance more revenue for you, then the solution provider coming to you should be telling you about that.”

AI also offers the ability to deploy ‘big data’, something that is ceaselessly talked about, but that has never fully delivered for its users. For contract caterers, this means integrating sales forecasts, event calendars and historical patterns to generate automatic staff rotas with the right start and finish times and skill mixes. It also provides the ability in real-time to re-evaluate and optimise bookings, weather and event changes.

“We've talked about big data for years,” says Moore. “Actually, no one ever uses it, because it's too costly to use, then too costly to interpret, then too costly to implement. But AI has changed that.”

Moore also talks of AI’s ability to deliver effective digital- and voice-led ordering systems. “AI has great voice ordering solutions,” he says. “I can now go to my phone and say, ‘Hi, I’d like to order a burger and chips now for 3.45pm please’.”

That ability can be integrated into existing apps and kiosks, enhancing speed and inclusivity in the ordering process. Again, an agentic model could involve one agent handling speech-to-text, another menu/rules, a third payments and loyalty – all co-ordinated by a top-level ‘concierge’ AI.

The possibilities for contract catering are endless, Moore maintains. They also include image generation of menus, something that is hugely important for younger generations, as he explains: “They want to see an image, understand it and buy it. They don’t want to read about it, they want to look at the plate and say, ‘That looks delicious’.”

Whether we’re talking about evolving gen z customer expectations, rising labour costs or fragile margins, Moore’s message to contract caterers is disarmingly simple: stop talking about AI and start talking about your problems. AI‑infused systems can already predict demand more accurately, schedule staff more intelligently and reduce waste more efficiently – and the next wave of agentic tools promises richer, more reliable insight still.

The real challenge, he argues, is not the technology itself, but operators’ ability to integrate it into everyday workflows. Those who manage that trick will quietly bank the productivity gains, while everyone else is still sitting in conferences arguing about what AI really means.

At the end of the day, Moore perceives AI as being less of a magic wand, more as a new layer in the plumbing of contract catering. Done properly, it accelerates development, unlocks data and nudges productivity in the right direction. Done badly, it produces expensive toys and unreliable numbers.

Most operators, he believes, are still somewhere in between: aware they need to act, but unsure how exactly to embed AI into the systems they already run. The opportunity is there – but it will belong to those prepared to ask hard, unfashionable questions about what technology is really doing for their people, their margins and their guests.

You may also be interested in…