Work that had to survive contact with real users.
Two AI products I build and run myself, a company brain built for a client, and advisory work inside large organisations. The products keep the advice honest; the advisory keeps the products grounded. Both are the reason I can tell you what will break before it does.
CaseStar
A focused practice room for aspiring consultants: realistic pressure, structured cases, and feedback that arrives while the habit is still fresh.
- My role
- Founder, architect and builder
- Built with
- AI agentsVoice UXReactSupabase
What running it taught me: people only trust AI feedback when it is specific, fast, and visibly grounded in what they actually said. That is the bar I hold client systems to.
- 012,000+ users
- 02Partnerships with top consulting firms
- 03Live voice coaching loop
Qube
Reads a business's bills, contracts and supplier terms and hands back the lines worth checking, so you can see where you pay more than you should without purchasing expertise. Nothing changes without your approval.
- My role
- Lead engineer
- Built with
- LangGraphData pipelinesNext.jsAI analysis
What building it taught me: when AI touches money, it is only as good as the audit trail it leaves and the approval it waits for. Legibility is not a feature; it is the product.
- 01Spend-cube generation
- 02Category insight
- 03Workflow-native AI
Zazi
A company brain for Zazi, the company founded by Jeanne de Kroon: one system that holds how the business runs and automates the operational work around it.
- My role
- Architect and builder
- Client
- Zazi, the company founded by Jeanne de Kroon
What it taught me: a company brain only earns trust when it holds how the business actually runs, not how the org chart says it does.
- 01Company brain
- 02Operational automation
Operational work that used to be done by hand now runs through one system the team can see and adjust.
Inside the organisations where AI has to answer to operations.
Alongside the three cases above, I have worked on strategy and delivery with teams at large organisations, usually in the room where executive ambition meets what the operation can actually absorb.
Most of the useful advisory work happens around a table, with the people who will run the system, not in a slide deck.
Heathrow
Strategy and sourcing
Buy-and-build strategy and tender guidance
Worked out which capabilities Heathrow should build itself and which it should buy, then guided the tender so the decision still held once suppliers, procurement and the people who would run the result were all in the room. At an airport a build-or-buy call outlives the project that made it, so the reasoning had to be right before it was fast.
What it showsBuild-or-buy judgment, running a tender, executive advisory in critical infrastructure
Rabobank
IT project management
IT project management for the mortgage application system
Ran the IT project management for Rabobank's mortgage application system: scope, plan, suppliers and stakeholders, in a bank where a mistake in the application process is a customer problem and a regulatory problem at the same time. A mortgage system sits where the bank's revenue meets its regulator, so the discipline of delivery mattered as much as the software.
What it showsDelivery leadership in a regulated bank, stakeholder management, keeping a critical system on track
CZ
Platform enablement
Set up the low-code platform that enabled development
Set up the low-code platform at CZ that let teams build and ship their own applications, with the guardrails to do it safely, so development no longer waited on one scarce engineering queue. For a health insurer that is the difference between an idea reaching the business and an idea waiting its turn.
What it showsPlatform set-up, enabling teams to build for themselves, governance that makes self-service safe
Also: Shell, Valcon, ING. Details in conversation.
If it needs to feel like a product, not a prototype, we should talk.
Replies within one working day. Or email yvoderooij@gmail.com.