Which data assets are ready?
What’s your “build vs buy” default?
Do you have a team that can carry a dish from idea to adoption?
Once adopted, are you clear how to support the entire chain, from data provision, to model, to outcome?
Most AI strategies read like an amateur restaurant’s kitchen renovation plans: glossy diagrams, shiny new gadgets and yet no meals coming out. You don’t need to stock a new kitchen with the latest technology to test a soup recipe. You need a clear menu, an efficiently stocked pantry and a way to get plates out of the door while the guests are still hungry.
Executives don’t ask for recipes, they ask: “will this be on the menu by December?” If your strategy can’t answer that in a sentence, you’re probably shopping for gadgets instead of planning a meal. Worse still, if you buy all those gadgets and try to use them all at once without a real plan, expect an awful result.
Most AI strategies stumble because they start with tools, not targets. A strategy that leads with “using large language models” is like a restaurant that brags about its oven but can’t tell you what’s being served.
The first move is deciding which outcomes matter. Choose a handful, three at most, that can move a business metric by year-end. Whether it’s revenue lift, cost reduction, risk avoidance, customer experience: whatever the organisation cares about most. Tie each one to a use case, with a named owner and a measure of success that your financial department would be able to show and sign off on.
This short menu keeps everyone honest. It forces you to say “no” to distractions and makes progress visible.
Before worrying about platforms or partners, write down a sentence for each of the three use cases: “We will move [metric X] by [Y%] for [process/customer Z] by [date D], with [use case U], led by [owner O].”
If you can’t do that, you don’t have a menu, just the ingredients on the counter.
Stock the Pantry, Staff the Kitchen
Once you know what’s on the menu, the question is simple: do you have the ingredients, the tools and the people to make it happen?
For AI, that means three things:
1. Data - Imagine data was an ingredient, let’s say tomatoes. You can buy them fresh, tinned, puréed, sun-dried etc. They are all still tomatoes but you can’t swap one for the other without consequence. What matters is when you buy them you know what you are getting, storing them properly so they’re always available for any recipe, not just tonight’s pasta. All too often, teams bring in data for a single use case and then leave it to rot. A good strategy makes sure every new batch of data goes into the shared larder, ready for future dishes. Teams love being able to reuse data that’s already been pulled in previously, it leads to faster turnaround times and much less work.
The trick is balance. Stock just enough data to make your recipes. Too much and you waste money, stock too little and you can’t deliver.
2. Architecture choices - Build, buy, partner. Not every organisation needs a custom kitchen, sometimes renting space in someone else’s works better. The key is interoperability, cost transparency and knowing your exit ramps so you don’t get trapped by one supplier.
3. People and skills - Who’s actually doing the cooking? You’ll need a mix: technical specialists, product managers and key stakeholders who connect business priorities to AI capabilities. Adoption champions inside business teams matter just as much, as they make sure the dishes actually get eaten.
Takeaway: Run a simple audit before you buy anything new:
If you can’t answer these, you’re still shopping, not cooking.
Cooking, Plating & Serving
Even with a good menu and solid ingredients, execution can collapse in the final stretch. In restaurants, that’s hygiene, timing and presentation. In AI, it’s supportability, governance, adoption and measurement.
Supportability - It’s one thing to create a standalone dish with carefully curated ingredients, a meticulously followed recipe and hours of attention to detail. It’s another to plate that same dish every single day, at scale, with the same quality and taste no matter who’s on shift. AI is no different. Everyone gets excited when a data scientist shows off a great outcome they have crafted on their laptop with hand-prepared data. But turning that demo into a daily service is a completely different test. Can it run flawlessly, end-to-end, with upstream data delivered on time and downstream teams actually using it? Who checks if the inputs are correct? Who fixes them when they’re not? Do models get re-run and who is responsible for watching them?
Governance and risk - Think of this as kitchen hygiene. Are there clear rules about safety, fairness and accountability? Who checks for model drift (when AI accuracy slowly decays), bias or security issues before anything reaches customers? What happens when the inspector visits? One bad dish can ruin the whole reputation.
Adoption - A dish isn’t finished when it leaves the kitchen, it has to be eaten. The same with AI. If teams don’t change their workflows, you don’t have value, you just have a demo. Track usage, reward early adopters and build feedback loops to improve.
Value tracking - Every strategy needs a receipt. Can you show, in numbers, that the finance department accepts that the investment is paying off? That means baselines and a ledger that records both benefits and costs.
Sustainability & Cost - Finally, the bill. AI can consume huge amounts of cost, energy and resources. A serious strategy looks at the compute, storage and service costs and sets targets so progress doesn’t leave a bigger footprint than the problem it solves.
Ask yourself:
Do we have a clear idea of the problems we are solving?
Can we support what we release?
Are people actually using what we have built?
Can we show demonstrable improvement?
Can the finance department trace the value on a ledger?
Do we know how our cost scales with usage?
If the answer is “no” to any of these, your kitchen isn’t ready for service.
Most companies stumble not because they lack ambition, but because they overcomplicate the strategy. They try to cook everything at once, buy every gadget or ignore the basics of hygiene. The result: a glossy presentation deck but a mess in the kitchen. The organisations that succeed keep it simple: achievable outcomes, the right ingredients and controls that make the food safe to serve. Everything else can wait.
So the question is this: if your AI strategy were a restaurant, would you be serving meals or bragging about the new range cooker and glass-door fridge?
About the authors
Larry is a lifelong technologist with a strong passion for problem-solving. With over a decade of trading experience and another decade of technical expertise within financial institutions, he has built, grown, and managed highly profitable businesses. Having witnessed both successful and unsuccessful projects, particularly in the banking sector, Larry brings a pragmatic and seasoned perspective to his work. Outside of his professional life, he enjoys Brazilian Jiu-Jitsu, climbing and solving cryptic crosswords.
LinkedIn
Ash is a strategy and operations professional with 14 years of experience in financial services, driven by a deep passion for technology. He has led teams and projects spanning full-scale technology builds to client-facing strategic initiatives. His motivation comes from connecting people, processes, data and ideas to create solutions that deliver real-world impact. Beyond work, Ash enjoys exploring different cultures through food and cocktails and practices yoga regularly.
LinkedIn