From Redundancy to Wine Tech: The Slightly Unlikely Story of SavvySipper

There are several ways to respond to being made redundant. You can spend six months furiously updating LinkedIn, convince yourself that drinking coffee while scrolling through job adverts counts as a productive morning, or become extremely interested in the phrase “open to opportunities”, despite having absolutely no idea what those opportunities might actually be.

I did all of those things to varying degrees. Then, after being brought in to consult for a leading cruise line to build a new B2B2C division, something rather unexpected happened. An idea that had been rattling around in my head started to take shape, and eventually I decided that, rather than simply looking for another job, I might have a go at building something myself.

So this is the rather unlikely story of how redundancy, a wine list and an unreasonable amount of time spent arguing with computers eventually became SavvySipper.

The Bit Where Everything Changed

At the beginning of 2025, I was made redundant. After years building a career in travel, following several years in entertainment, it wasn't exactly the plot twist I had envisioned. Professionally, things had been going rather well until, rather suddenly, they weren't.

I won't bore you with the corporate details. Partly because I had a six-month non-compete, partly because there are always several sides to these stories and partly because nobody needs another overweight middle-aged bloke on the internet writing a thinly disguised corporate memoir.

Redundancy does, however, give you something most busy executives don't get much of: time. Time to think about what you actually want to do next and, perhaps more importantly, whether you really fancy spending the next ten years climbing another corporate ladder.

One thought kept coming back to me: what if I built something myself?

That last bit turned out to be rather important.

It Started With a Wine List

The idea for SavvySipper came later, during a business meeting at Quill Wine Bar in London Bridge in September 2025.

While waiting for my client to arrive, I started looking properly at the wine list. One thing in particular caught my attention: the pricing of the different serving sizes. The 500ml carafe was considerably better value than the 750ml bottle.

Most people would probably have ordered whichever sounded good, asked their friends what they fancied and got on with dinner. I started doing the maths, because apparently this is what happens when you have too much time on your hands.

The more I looked, the more interesting it became. Restaurant wine pricing isn't simply about what the restaurant paid for the bottle. There are different serving sizes, glass prices, bottle prices, retail benchmarks, perceived value and a whole collection of psychological pricing decisions hiding behind what looks like a fairly innocent wine list.

And that raised a question that bothered me considerably more than it probably should have: why can't someone just tell you which wines are actually worth looking at?

Not after taking a three-year course in wine terminology. Not after subscribing to some extraordinarily serious wine service. And not after asking a sommelier whether a £48 bottle represents good value, only to receive an answer involving gravelly soils, elevation and somebody's uncle in Burgundy.

What if technology could simply do the hard work?

That was the beginning of SavvySipper.

From Wine List to Wine Intelligence

The original idea was deliberately simple. Take a restaurant wine list, read it, identify the wines, understand the serving sizes, compare the prices with real-world retail pricing and work out which bottles, glasses and carafes looked interesting. Then add useful food pairings, wine information and some sommelier insight, and turn all that rather tedious information into something a normal human being could actually understand.

The idea was never to tell people what wine they should drink. If you want to spend £200 on a bottle of Bordeaux, crack on. The point is to give you enough information to decide whether you want to spend £200 on it.

Drink what you like, understand what you're paying for and, ideally, avoid discovering afterwards that the £22 bottle you could have bought elsewhere has just cost you £78 in a restaurant.

That thinking eventually became the foundation of SavvySipper's SavvyScore and wine-list analysis.

Then I Had to Actually Build the Thing

There was only one fairly significant problem. I had absolutely no idea how to build software.

I understood the business idea. I understood the wine problem. I understood the customer. What I couldn't do was actually make the bloody thing.

So I started learning. Coding, databases, APIs, GitHub, deployment, email infrastructure and all sorts of other things that previously lived in the mysterious world of “IT”, where I had always assumed somebody else knew what they were doing.

I learned to work with GitHub, deploy applications using Vercel, build database infrastructure with Supabase and deal with email infrastructure using Resend.

Then AI-assisted development arrived and changed the rules again. Tools such as Google Antigravity became part of the process, giving me something somewhere between a coding assistant, technical sounding board and extraordinarily patient colleague who never appears to get bored of the same question being asked for the seventh time.

It was extraordinary. It was also occasionally maddening. You can spend 45 minutes explaining exactly what you want, press the button and then watch it confidently do something completely different.

Progress, apparently.

The Development Chat History Should Probably Be Classified

If you read through some of my development conversations from the past year, you might reasonably conclude that I either became a software engineer or suffered a fairly serious breakdown.

There have been questions like “Why has this deployment suddenly stopped working?”, “Why is the database returning the wrong wine?”, “Why has fixing this button broken three other buttons?” and “Why does the mobile version look like that?” There was also the occasional discussion about why a CSS card grid was behaving like a stroppy teenager.

And then there are the particularly productive days when you stare at the screen for a while and eventually ask: “What the hell have I just done?”

There is a particular joy in building something yourself. Every problem is yours. Unfortunately, every problem is yours.

AI Didn't Build SavvySipper For Me

There is a slightly lazy idea that AI-assisted development means you type a sentence, make a cup of tea and return to find you've accidentally built the next Microsoft.

That hasn't been my experience.

AI has accelerated my learning enormously. It has helped me understand code, debug problems, structure databases, investigate deployment errors and attempt things I probably wouldn't have touched without it. It has also made it possible to explore ideas that might otherwise have remained just that: ideas.

But it doesn't remove the need to understand what you've built. When something breaks, saying “the AI did it” isn't particularly useful. The website is still broken, the database is still broken and the wine scanner isn't going to magically fix itself because you've stared at the screen angrily.

Trust me. I've tried.

Wine Lists Are Surprisingly Difficult Things

The technology became particularly interesting once I started trying to make the wine scanner genuinely useful. Restaurant wine lists are wonderfully inconsistent documents. They have different fonts, different layouts, different vintages, different bottle sizes, different spellings and different ways of describing exactly the same wine.

And occasionally you get a photograph taken at an angle that appears to have been captured during an earthquake.

Reading the words is one thing. Understanding what those words actually represent is another. A wine list might give you a producer, grape, appellation, vintage, bottle size and price, or it might give you four of those things and apparently expect you to work out the rest through intuition.

That meant SavvySipper had to become more than an OCR tool. It needed to recognise wines, normalise names, match producers and vintages, compare prices and turn a fairly messy collection of menu data into something genuinely useful.

In other words, the wine list needed a translator.

What Does SavvySipper Actually Look At?

As the platform developed, the idea evolved into a broader system for understanding the economics and quality of a restaurant wine list.

Margin Fairness looks at restaurant pricing against wider retail value. Crowd Score brings consumer sentiment into the equation. Glass Tax looks at the economics of ordering wine by the glass rather than the bottle, while Menu Discovery considers the breadth and variety of the list. Then there are the Sommelier Insights, which are designed to turn all those numbers into something resembling useful wine advice rather than an accounting spreadsheet.

Because a wine list isn't simply “cheap” or “expensive”. A restaurant can have expensive wines that are genuinely excellent value, inexpensive wines that aren't particularly exciting, or a brilliantly curated list with slightly bonkers glass pricing.

It can also have a fantastic bottle sitting quietly on page seven while everyone orders the second-cheapest wine because nobody wants to be responsible for making the wrong decision.

That is the bit SavvySipper is designed to uncover.

It Turns Out This Isn't Really Just About Wine

The deeper I got into the project, the more I realised that the underlying problem isn't really wine. It's information.

Restaurants have information. Retailers have information. Producers have information. Wholesalers have information. But the person sitting at the table often has remarkably little. They have a list, some prices, perhaps a vague understanding of a few grape varieties and their dining companions looking at them as though they have just been appointed chairman of the wine committee.

So why not give them better information?

That's really what SavvySipper is about. Not telling you what to drink, but helping you understand what is in front of you. Finding the interesting bottles, spotting better-value pours, understanding restaurant mark-ups and making an intimidating wine list considerably less intimidating.

I'm Still Learning. Obviously.

The slightly ridiculous thing is that SavvySipper is still evolving, and I'm not sure it will ever really be finished. Every new feature creates another question, every new data source creates another problem and every improvement to the scanner seems to reveal another wonderfully obscure edge case hiding somewhere in the database wearing a tiny disguise.

And every time I think I've finally understood the technology stack, someone invents another technology stack.

Wonderful.

Exactly what I needed.

But that is also what I wanted when I started this. After a pretty brutal period professionally, I wanted something that gave me a reason to keep learning, experimenting and building.

I've learned more about software development, databases, AI, deployment, APIs, email infrastructure and product design over the past year than I could ever have imagined when this started. Most of it because something was broken, which is apparently a perfectly legitimate educational strategy.

So Why SavvySipper?

Because wine lists shouldn't require a degree in pricing psychology to understand. There are some brilliant wines hiding on restaurant menus that most people will never discover, and understanding what you're actually paying for before ordering a bottle seems like a fairly reasonable thing to want to do.

Because restaurants deserve better data about their wine lists. Because retailers and producers should benefit when someone discovers a wine they genuinely love. And because drinking wine should be enjoyable rather than intimidating.

Most of all, because there is something enormously satisfying about taking an idea that started with a slightly odd-looking carafe price in a London wine bar and turning it into something real.

The Savvy Verdict

SavvySipper started with a wine list, a pricing anomaly and one fairly persistent question: why doesn't this make sense?

That question eventually turned into a much bigger one: what if technology could make restaurant wine lists easier to understand?

That's what I'm building.

It has involved wine, data, OCR, AI, databases, GitHub, Vercel, Supabase, Resend, Google Antigravity, an unreasonable amount of caffeine and more conversations with coding assistants than I would care to admit publicly. It has also involved a fair amount of swearing.

But we're getting there.

If you'd like to see where the experiment goes next, keep an eye on The Pour. There will be wine, there will be data, there will probably be another technical catastrophe and, hopefully, considerably more bottles where the price actually makes sense.

Cheers.

Stephen
Founder, SavvySipper