Let’s get straight into it… I’m going to start sharing the AI strategy for hospitality and AI optimisation work we’ve been doing at Brew.
We’ve spent the last two years watching, testing, breaking, and rebuilding. Not theorising, not reposting industry takes – actually using the tools, building with them, and figuring out where they hold up and where they don’t.
2026 is the first year we’ve been confident enough to put AI tools in front of the team for real client work.
This isn’t a sales pitch for AI. It’s a working account of what we’ve learned so far.
The short version: AI is useful, but only when it’s kept on a leash.
Article summary
- AI strategy for hospitality: The perception problem
- Where people still fit in
- Brew’s new AI-related platforms
- The subscription vs API question
- The security issue nobody’s talking about enough
- AI strategy for hospitality: Where is this going?
AI strategy for hospitality: The perception problem
The general sentiment around AI hasn’t been positive, and that’s not irrational.
Most businesses were first exposed to models that were inconsistent, overconfident, and often wrong. Layer on top of that the constant narrative around job replacement, and it’s easy to see why adoption has been hesitant or reactive.
We’ve felt that internally as well. Perception shapes rollout more than the technology does. If the first experience people have is poor, that sticks.
That’s a big part of why we’ve moved deliberately. We’ve been tracking AI closely since the early OpenAI betas, but we held back on full rollout until the outputs reached a level where they could genuinely support client work.
Even now, the focus is balanced usage. Not offloading too much to AI. Not forcing it into places it doesn’t belong. Making sure that at every step, it’s adding real value rather than just speeding something up for the sake of it.
From a client perspective, the goal isn’t to reduce effort and do the bare minimum. It’s the opposite. AI gives us the ability to do more within the same constraints – deeper research, more considered creative, more complex builds. The output is a starting point at best in most cases, not the finished product.
Where people still fit in
Two years ago, when AI imagery started to become viable, we made a decision that went against the direction of travel. We invested in professional photography.
We bought the kit, built the capability in-house, and made it a core part of what we offer. It’s now one of the fastest-growing parts of the business.
That decision shaped everything that came after.
Hospitality is specific. Guests book based on what they see. A hotel room, a plate of food, a bar interior – these aren’t abstract concepts. Someone is going to walk into that space and expect it to match what they were shown.
Generic AI imagery doesn’t hold up to that. And increasingly, audiences are good at spotting it.
AI-generated assets are now the baseline. If you want to stand out, you need to add the thing AI can’t replicate. That’s where people come in.
For us, that means real photography, client-specific knowledge, and human review at the right points in the process. AI supports that, but it doesn’t replace it.
Brew’s new AI-related platforms
Off the back of that thinking, we’ve been building two internal products: Citewatch and Talos.
Both started as rough ideas. Both have been rebuilt multiple times as the technology and our understanding have improved. Both are still early.
Citewatch
Citewatch is a Generative Optimisation (GEO) tracking tool.
It measures how businesses show up in AI-generated answers. Not rankings in the traditional sense, but presence, visibility, and how often a brand is referenced or cited.
This uses tokenised questions to create large amounts of questions that are sent to the AI systems, and it builds a “share of voice” and a weighted version called a Citewatch Score. This enables you to see where the gaps are in your digital marketing, along with when and why your business shows up in those responses.
Talos
Talos is a content creation engine, but not in the generic sense.
At its core, Talos is built around a client’s own assets. It starts with a media library that holds original photography. From there, you can layer in optional vision AI tagging and descriptions, or manage it manually if preferred.
The second layer is the knowledge base. This is where web pages, documents, and research queries are stored and continuously updated. Instead of generating from scratch, the system pulls from this base using semantic search, which keeps the output aligned with the client’s tone, offering, and identity.
The third layer is templating. Rather than open-ended generation, we build structured templates that AI populates. This keeps output consistent, on-brand, and usable. It also allows for manual control where needed – AI is optional, not forced.
This matters in hospitality. A restaurant group or hotel brand has a distinct feel. The food, the interiors, the atmosphere – generic AI output flattens that. A knowledge-led approach preserves it.
There’s also a practical angle. The current “stack” approach – multiple subscriptions stitched together – isn’t sustainable. When one custom tool can do the same job, more efficiently and with better output, that model starts to break down.
AI strategy & tools: Subscription vs API
This is the part that doesn’t get talked about enough.
When you start rolling AI out internally, the first question is usually: who needs a subscription?
Then it becomes: which level?
Then: what happens when people hit usage limits?
If you build workflows that are actually useful day-to-day, standard subscriptions will hit caps. At that point, you start layering in API usage to top up. Then you remember that pricing is still subsidised, and usage limits are tightening across providers.
The reality is that most people won’t get enough value from a standalone AI subscription to justify it. A smaller group will – the power users.
Our approach is shifting towards those power users building tools for everyone else.
That means focusing on model and token efficiency. Using cheaper models where they’re good enough. Refining prompts instead of brute-forcing results. Designing tools around specific workflows rather than general use.
The wider team then uses those tools, without needing to understand the underlying complexity.
This keeps costs predictable, reduces subscription sprawl, and improves output quality. It also avoids being locked into a single provider, which is going to matter more over time.
The AI security issue nobody’s talking about
Security is probably the biggest overlooked issue in AI adoption right now.
There’s a tendency to connect everything to everything and see what happens. That’s not going to end well.
The UK Cyber Resilience Act is coming this year.
Supply chain security checks are increasing. At the same time, businesses are rolling out AI systems with broad, often unchecked access to internal data. That’s a problem.
We wrote our AI policy early and have been strict about what these systems can and can’t connect to. Not everything needs to be accessible. In many cases, it shouldn’t be.
For most businesses, the hardest part of AI rollout won’t be the technology. It’ll be using it securely.
AI strategy for hospitality: Where is this going?
AI isn’t going away, but the way it’s being adopted right now will change.
The competitive edge won’t come from having the most subscriptions or the latest tools. It’ll come from building the right things, keeping humans in the loop where it matters, and approaching the whole space with a level of scepticism.
Used properly, AI lets you do more, with more consistency, and often at a higher standard. Used poorly, it creates noise, risk, and a lot of wasted spend.
We’re still figuring this out as we go, and we’ll be sharing our learnings – and mistakes, one of humans’ charming abilities – as we go.
Get in touch: Let’s create your hospitality AI strategy
Let’s chat about what AI tools and strategies your business could, or should, be using.
If you’re wondering how your business is showing up in AI-generated answers and search engines like ChatGPT, let’s talk about Citewatch.
If you need a helping hand with your content creation – from image alt tags and knowledge organisation to structured templates for future content, let’s get into Talos.
AI strategy for hospitality: FAQs
Not universally. The value depends on how it’s applied. In most cases, AI is best used to support specific workflows rather than as a blanket solution.
On its own, rarely. It can help with structure and speed, but without real imagery and client-specific input, it tends to feel generic.
Usually, not across the board. A smaller number of power users combined with shared, purpose-built tools is often more effective and more cost-efficient.
Yes, if done without structure and control. Access should be limited and intentional, with a clear policy in place.
Start with a single workflow where AI can add clear value. Test it, refine it, and build from there rather than rolling it out everywhere at once.
Adam Cox
Adam is Brew’s technical director, and he keeps our agency at the forefront of tech in hospitality marketing, from client-pleasing platforms to efficiency-driving automation.
