Founder · GenAI · Bootstrapped to Profitability · UK / US / AU
I identified an inflection point in generative media before it became obvious to the category: once mask-based image editing, inpainting, and face-preserving generation became good enough, personalized children’s books no longer had to be a manual Photoshop workflow. They could become a scalable software-and-operations system.
Around that shift, I built Fableau: a personalized story product company that combined GenAI, editorial production, ecommerce, print-on-demand infrastructure, and performance marketing into one operating model.
The result was not just a demo — it became a real commercial system with shipped products, repeatable content production, working acquisition channels, and early market convergence in the UK.
Upload a child’s photo → choose a story → AI generates a fully personalized illustrated book → printed and delivered as a premium physical product. Not a template with a name swap — every illustration features the child as the main character.
The real breakthrough was not “AI writes stories.” The real breakthrough was that image personalization crossed the threshold from novelty into manufacturable quality.
Once it became possible to take a child’s photo, preserve facial identity with masks, inpainting and face-swap workflows, and generate illustrations consistently enough for a printed product — a new type of company became possible. Not a design studio, not a traditional publisher, not a pure AI image app. An AI-native story product company.
The bottleneck shifted from “can we technically make it?” to: can we orchestrate reliably, turn outputs into premium products, ship fast, create enough SKUs, and acquire customers profitably?
Five layers of a venture, built simultaneously.
Photo upload → face swap → inpainting → generation → rendering → layout → print-ready PDF. Queue-based processing with RabbitMQ, worker pools, fault tolerance, and observability via Grafana.
Writers, editors, illustrators, designers, ML engineers, DevOps, QA. Product quality came from the interaction between editorial taste and the AI pipeline. Automation where it improved throughput — humans where it had to feel gift-worthy.
Web-Editor 1.0 reduced dependence on engineering for every new title. New book launch cycle cut to ≤5 days, development speed 3x. Changed the company from “a team that can make books” into “a system that can release and test books quickly.”
Full storefront, purchase flow, preview experience, upsell logic, digital add-ons (PDF, video greetings). Print API handoff to production partners. Cross-border fulfillment across 30+ countries. This made the product buyable, not just generatable.
UK-first rollout via Meta + Google Ads. Emotional UGC outperformed technical process explanation. The pattern: brute-force learning → creative-market fit → channel discipline → scale under seasonal demand.
Sep 2025
0.6x
ROAS · CPA £66
Oct 2025
1.6x
ROAS · CPA £24
Nov 2025
2.5x
ROAS · CPA £28
Dec 2025
1.9x
ROAS · CPA £30
Key learning: Meta CPA improved from £55 → £23 in one month. Google CPA from £204 → £14 over 4 months. The winning creative: a New Year’s Book UGC video — emotional resonance, not technical novelty, was the growth lever.
Real feedback from parents who bought personalized books.
This project is a representation of how I work: I look for enabling technology shifts with commercial consequences. I translate them into product theses, not abstract trend commentary. I build across product, GTM, content, and operations together. I care about the whole system — demand, margins, fulfillment, retention, and speed of iteration.
Let’s talk about venture building, GenAI products, or turning technology shifts into businesses.