Most retailers already own their best salesperson, and most of them keep it locked in a spreadsheet. A store with 100,000 products has 100,000 chances to describe, persuade and get found - yet the fields sit half-empty, copied from a supplier feed, or written once in 2019 and never touched again. Ocula Technologies, a company from Belfast, decided that was a problem worth building an entire AI stack around.
Founded in 2021 by Thomas McKenna, Gregory Fletcher and Gerry Buggy, Ocula builds software that reads a retailer's catalogue, fills in what's missing, and rewrites product listings to be clearer, more on-brand, and easier for search engines to surface. The company's early product, Ocula Boost, scored product pages against best-in-class competitors and generated optimized titles, descriptions and image alt-text. The pitch was blunt: a five-minute setup, no IT involvement, and content produced at a speed no human copy team could match.
What has made Ocula more interesting than a faster copywriter is the shift happening underneath ecommerce itself. For two decades, being found online meant ranking on Google. Increasingly, people ask a machine - ChatGPT, Perplexity, Claude - what to buy, and the machine answers from whatever product content it can parse. Ocula reframed its whole thesis around that moment. Its internal shorthand for it is disarmingly plain: when machines go shopping.
The catalogue nobody has time to fix
The problem Ocula solves is unglamorous, which is part of why it's valuable. Product data is messy. Sizes are missing, materials are blank, two suppliers describe the same category three different ways, and the marketing team that could fix it has a few writers and a few hundred thousand items. Manually, the maths never works. Ocula's own framing puts the cost of doing it by hand at more than £250,000 for every 10,000 SKUs - the kind of number that explains why so many product pages stay broken.
Rather than sell a single writing tool, Ocula built a set of specialized AI agents that split the job the way a real content team would. One enriches, one writes, one checks the work, and an orchestrator keeps them in step. It is a deliberately different bet from pasting a catalogue into a general chatbot.
How the agents divide the workEnrichment Agent
Fills missing product fields and cleans up incomplete supplier feeds.
Copywriter Agent
Generates on-brand, SEO-tuned descriptions at catalogue scale.
QA Agent
Grades output for accuracy and brand consistency before it ships.
Orchestrator
Coordinates the agents and pushes content across channels.
The QA step is the detail worth pausing on. Ocula's system effectively has one agent grading another's homework before anything goes live - an admission, built into the product, that generated text needs a check. A separate performance agent then watches how the published listings actually do across search and AI channels, closing the loop.
In practice, a merchandiser doesn't sit and write. They point Ocula at a catalogue, set the brand tone and the rules, and let the agents work through the backlog - a category at a time, tens of thousands of items in the time a small team would clear a few hundred. The output isn't a first draft to be rewritten; it's meant to publish, with the QA agent flagging anything off-brand or inaccurate before it does. That changes what a content team spends its day on: less typing the same size-and-material sentence for the thousandth time, more deciding voice, priorities and which channels matter. The company's origin story leans on this exact frustration - its founders watched retailers sit on their most valuable asset, product data, while vendor pilots and in-house AI proofs-of-concept stalled.
Optimizing for readers who aren't human
The phrase circulating in retail circles is Answer Engine Optimization, or AEO - the successor to SEO for a world where a chatbot, not a results page, gives the answer. When a shopper asks an assistant for "a reliable washing machine under £400," the assistant reads structured, well-written product content and recommends from it. If your listing is thin, you don't rank low - you simply don't appear. Ocula tunes content for both the old channels and the new ones, which is a large part of why retailers with enormous catalogues have taken the meeting.
The distinction from a general chatbot matters here. Anyone can ask ChatGPT to write a product description; few can get it to write half a million of them, on brand, without drifting into invented specifications, and then verify each one. That is the gap Ocula is built for - not the clever sentence, but the reliable factory that produces clean, consistent, checkable content at the scale a real catalogue demands. The specialization is the point: narrow agents doing narrow jobs, measured against traffic and conversion rather than vibes.
Who is actually using it
Ocula's client list reads like a walk through a British high street and an American strip mall. Electronics retailer AO, Walgreens Boots Alliance's Boots, model-maker Hornby, Asda, Ryman, CeX and second-hand specialists sit alongside US names such as Five Below, Living Spaces and On Running, plus Blain's Farm & Fleet, The White Company and Vintage Tub & Bath. These are catalogues measured in the tens or hundreds of thousands of items - exactly the scale where manual editing stops being an option and Ocula's economics start to make sense.
The business model is straightforward B2B SaaS: retailers subscribe, priced against the scale of their catalogue and the modules they use, and Ocula frames its value against the copywriting costs it removes and the organic traffic and conversion it adds. Michael Manton of Hornby and Clare Evans of AO are among the client voices the company points to.
Belfast, and the cheques that back it
Ocula chose Northern Ireland for its software development centre, committing £11m to research and development and building its team around a Belfast base. That decision came with roughly £5m of strategic investment from Lloyds Banking Group and Castelnau Group, and support from Invest Northern Ireland. In July 2024 the company added a £3.25m Series A led by Praetura Ventures, with Castelnau and Lloyds returning - funding it framed as a plan to treble in size.
Castelnau, an investment vehicle in the orbit of Phoenix Asset Management, holds a sizeable stake - around 41% by its own portfolio disclosure - and lists Ocula in its "AI & Technology" category. An earlier round valued the company near £10m. It is a real, revenue-generating retail-tech business rather than a demo-day curiosity, and its backers keep re-upping.
The people and the posture
The team is small and technical - around 26 people spanning product, data science and engineering, led by McKenna as CEO and Fletcher as CTO, with co-founder Gerry Buggy (previously founder of data-privacy company Privitar) now advising. Ocula states four values: Be Brave, Empower Others, Seek Growth, and Spark Joy - the last defined as building things people will love and leaving them better than you found them. It's a lighter register than the average enterprise-software mission statement, and it fits a company whose product is, at its core, about words.
The values also hint at how Ocula wants to be judged. "Seek Growth" and "Be Brave" are common enough on a careers page, but "Spark Joy" - defined as removing friction and small acts of thoughtfulness - is an unusual thing to ask of software that mostly rewrites bullet points about vacuum cleaners. It reads as a bet that the dull work is worth doing well, and that the retailers who notice are the ones who stay.
Where Ocula sits in the market is still being drawn. On one side are general AI writing tools and the temptation to just prompt ChatGPT directly; on the other, product-information and analytics platforms, and much larger data players. Ocula's wedge is the specific, measurable, catalogue-scale job of product content built for the age of AI search - and a QA layer that keeps a human-shaped standard on machine-written text. If the way people shop really is shifting from searching to asking, the companies that wrote their listings for humans will need someone to translate. Ocula is volunteering.