AI transformation
AI, in practice.
Most AI pages are a promise. This one is an inventory: the vendors I stopped paying, the product I built with agents, the processes it runs, and the method for doing the same inside an organisation that isn't mine.
01 · What it does for me
It does part of the work. Not all of it. Part of it, every day.
At QS Ventures I can't name a process AI doesn't touch. Sales pipeline analytics run on it end to end. Agents handle the recurring work. Analysis, drafting and research go through Claude and Gemini before they reach me. This site was built from scratch the same way, including the assistant you can talk to.
The clearest example is flip.codes, an anti-fraud ticketing product I designed and built with AI coding agents: the credential protocol, the backend, the field-test tooling, and an operations layer of 21 autonomous-agent routines that run the business day to day, with one human review loop, daily. I am that human.
“Things I would have sold for $500,000 a couple of years ago, I can now build myself in two months for a little over $1,000.”
That changed what I think an operator is for. Not to buy the software. To know which parts of the company no longer need a vendor, a process or a headcount, and to be the person accountable when the agents get it wrong.
02 · Vendors replaced
Build what used to be bought.
- Project management: we ran the company on monday.com. We replaced it with software we wrote ourselves, shaped around how we actually operate instead of the other way round.
- Websites: ours lived on Framer. Now we build them from scratch, this one included.
- The calculation flipped. Buying used to be the cheap, safe choice. Now the question for any company is not which tool to buy next. It is which of the tools it already pays for it no longer needs.
03 · Method
AInabled: how I take AI into a company that isn't mine.
Most companies invest in AI and most of those investments underperform. It is almost never the technology. It is a pilot with no strategy behind it, a tool that doesn't talk to the systems the business runs on, staff quietly using whatever they found, or a blanket ban that stops anything from being integrated at all. AInabled is the service we built at QS Ventures to fix exactly that, in five stages and twenty tracks.
01
Know where you stand
Business analysis, AI maturity and capability assessment, and a hard look at the data and the compliance position (GDPR, AI Act). It ends in a readiness report the board can make a decision on, not a slide about potential.
02
Decide what, and with what
Use cases ranked by feasibility and return. Tools chosen against the systems already in place, not against a vendor's demo. Workflows that say which agent works with which team, and who signs off.
03
Deploy, agents first
Pilots with success criteria and KPIs written down before they start. A reusable portfolio of agents by business function. Integration with the systems that matter. And change management, because a tool nobody adopts is a cost.
04
Measure and tune
Dashboards and performance reviews. Bias and compliance audits. An impact framework with financial and operational benchmarks, so the conversation is about numbers, and a plan for scaling what works.
05
Make it stick
Governance and risk frameworks, the AI roles the company needs to hire or grow, a playbook so the knowledge doesn't leave with the consultants, and a roadmap for the next phase. Reference point: seven months for a 500-person company.
AI adoption service
AInabled
Assess, design, implement, optimise: a five-group, twenty-track framework for taking AI into an organisation, agentic pilots included.
Visit site ↗
04 · Built on it
Agents, voice, and two retail pilots.
Anti-fraud ticketing
flip.codes
Rotating, device-bound cryptographic credentials against ticketing fraud. Designed and built with AI coding agents: protocol, backend, field-test tooling and an operations layer of 21 autonomous-agent routines with a daily human review loop.
Visit site ↗
Voice studio
Divina.Machina
Art studio using technology to research how people interact with voice interfaces. Voice agents and voice branding.
Visit site ↗Marketplace
Agent-Place
Multi-vendor marketplace for prebuilt automations and AI agents. We built the marketplace, not the agents.
Earlier, at TheCUBE, I built AI-in-retail pilots that produced MVPs for two autonomous-store concepts, alongside a 14-country e-business roadmap for a €3B food group. Agent-Place, a marketplace for prebuilt agents, has since closed. It stays on this page because I don't list only the ones that worked.