An interactive case on why AI pilots stall
By Dr. Pablo Ramos-Sánchez, Product Lead at Arquestra. PhD in Chemistry, KU Leuven × Covestro, industry-funded.
A short playable case. You run an AI pilot for a chemical client over 30 days with a small team of three juniors. You choose what they learn (building AI, data architecture, protecting data, shipping reliably, organizing the team, talking to clients) and how to answer the client's requests. Shortcuts save days now and cost trust later. Halfway through, a steering review decides whether the pilot scales or waits, depending on whether the MVP shipped, whether the data question was answered, and whether the client trusts the team. There is no single winning path and there is not time for everything.
The point the case makes: most AI pilots do not fail, they wait. They wait at the point of trust (security, compliance, reliability, contracts), and they move again when someone owns the road from demo to production.
- Insights: why products get stuck, and who should own them (a three-minute read with the data)
- Arquestra, the product
- Who I am