Scaling industrial AI starts with deciding where machine authority ends and human judgement begins. In his thought leadership presentation at Future Digital Twin & AI Amsterdam 2026, Pedro Alcantara, EVP & CRO at Falkor, argues that organisations need to build trust in AI supported decisions before giving systems greater autonomy.
Using a compressor failure prediction as an example, Pedro distinguishes between detecting a signal and deciding what to do about it. AI can identify anomalies and rank maintenance priorities, but choices involving safety, production and commercial consequences may still require human judgement. Each decision boundary should reflect the evidence available, the operating conditions and the risks involved. As confidence grows, that boundary can move within clearly defined limits.
Pedro explains that trustworthy decisions depend on context. Industrial information is often spread across systems, documents and the knowledge of experienced staff. Connecting that information around assets and workflows gives people and AI a shared view of the operation. He also describes how bringing relevant context into day-to-day maintenance work can make recommendations and actions easier to assess and trace.
The presentation closes with a practical route towards greater autonomy: connect operational knowledge, define how work is performed, make decisions traceable and expand AI’s authority only when the evidence supports it.
Speaker
- Pedro Alcântara, EVP & CRO, Falkor
