AI-assisted ordering for construction
An industrial customer workflow, shaped around adoption. At Holcim, I led strategy and delivery for an AI-assisted ordering experience. The Spanish pilot is discussed in a published McKinsey interview. This case separates my leadership contribution, reported pilot outcomes and the decisions another business would need to assess before scaling.
Evidence and scope
Public case + published pilot results
Strategy and delivery leadership within the Holcim team
Spanish pilot with construction customers. Figures as publicly reported in the interview for that pilot scope. They are not an independent audit, a benchmark, or a guaranteed outcome elsewhere.
Outcome and evidence
An industrial customer workflow, shaped around adoption. At Holcim, I led strategy and delivery for an AI-assisted ordering experience. The Spanish pilot is discussed in a published McKinsey interview. This case separates my leadership contribution, reported pilot outcomes and the decisions another business would need to assess before scaling.
My role
My contribution was strategy and delivery leadership within Holcim's team: connecting the commercial use case with the customer experience and enterprise workflow. McKinsey identifies my role at the time as digital manager of global sales excellence.
The problem
Customers ordered cement by phone. The pilot tested whether a familiar messaging channel could make that routine easier.
The approach
WhatsApp was the customer interface. AI used order history to propose an order; customers could adjust and confirm it. Delivery required coordination across business and technology partners.
Constraints and governance
Testing exposed inappropriate product suggestions and interaction timing issues. Human escalation remained available. These are operating design questions, rather than proof of unrestricted autonomy.
Outcome and evidence
- Digital ordering adoption reported moving from 25% to 93% in the pilot.
- Around 66% of initial order proposals generated by the assistant were accepted.
- Figures as publicly reported in the interview for that pilot scope. They are not an independent audit, a benchmark, or a guaranteed outcome elsewhere.
- Revenue, retention and global-rollout outcomes are not quantified here. Future plans discussed in 2025 are not presented as completed delivery.
Outcome and evidence
- Digital ordering adoption (reported, Spanish pilot): 25% → 93%
- Initial order proposals accepted (reported, Spanish pilot): ~66%
- Source: McKinsey interview, Feb 2025
Evidence and scope
Spanish pilot with construction customers
Figures as publicly reported in the interview for that pilot scope. They are not an independent audit, a benchmark, or a guaranteed outcome elsewhere.
- Digital ordering adoption reported moving from 25% to 93% in the pilot.
- Around 66% of initial order proposals generated by the assistant were accepted.
Sources
Canonical URL: https://juanbeltran.ch/portfolio/worlds-first-autonomous-agentic-ai-in-construction