HVAC systems · Machinery & Manufacturing · IoT and AI

An AI-enabled IoT platform for thousands of installations

A control room monitoring thousands of refrigeration installations had to grow 30% a year without growing its staff at the same pace. We designed the platform and the operating model with AI.

Challenge: Connected installations growing 30% a year and a control room that could not grow at the same pace.
What we did: Operator shadowing, machine-learning alarm prioritisation and an AI assistant to resolve cases.
Result: Growth without a proportional increase in staff, onboarding from 3–6 months to 4–6 weeks.

+30%

connected installations a year, handled without a proportional increase in staff

4–6 weeks

to train an operator, instead of 3–6 months

The context

A leader in industrial and commercial HVAC systems, with a central control room monitoring thousands of refrigeration installations for food, pharma and retail customers.

The existing platform handled the historical installed base, but with connected installations growing 30% a year the digital infrastructure had to evolve.

The problem

Operating costs grew with the number of installations: scaling the control room meant hiring proportionally. Operators were flooded with alarms, all treated with the same urgency.

Training a new operator took three to six months, and turnover dispersed the know-how of the most experienced. Management was purely reactive: problems could not be anticipated.

The NiEW method

Designing the operating model, not just the technology

01. Business discovery

Analysis of shifts, workloads, escalations and control room economics; growth scenarios and success KPIs.

02. Operator research

Shadowing in the control room, interviews with senior and junior operators and supervisors, analysis of alarm history and tacit knowledge.

03. Architecture and prototypes

Data architecture for the machine-learning models, LLM integration and iterative tests with operators to validate suggestions and usability.

The solution

A platform where AI acts as filter and guide: it shows operators only what matters and suggests how to solve it, with the experience of their most expert colleagues.

Smart alarm prioritisation by severity, urgency and business impact, with a noise-free dashboard.
AI resolution assistant: paths suggested from similar cases solved in the past and troubleshooting guides generated in natural language.
Built-in onboarding with real cases and simulated training before handling live situations.
Knowledge retention: senior operators’ knowledge stays available to the whole team.

The results

Sustainable growth

30% more installations every year without a proportional increase in staff.

Faster onboarding

From 3–6 months to 4–6 weeks for new operators, thanks to the AI assistant.

Turnover under control

Knowledge stays in the system even when experienced people leave.

A foundation for growth

The platform can evolve into a simulator for training on critical cases, with no risk to real installations.

More case studies

Machinery & Manufacturing

A customer portal that answers before the customer calls

Challenge
Customers all over the world and a service team flooded with repetitive questions.

What we did
Needs verified with customers and the service team, a portal with an AI assistant on the documentation, code handed over to the internal team.

Result
A single point of access for orders, spare parts and documentation, and a service team free to focus on the cases that matter.
Machinery & Manufacturing

Configure, price, sell: the CPQ platform

Challenge
Highly configurable products, six weeks to qualify a request and deliveries of up to six months.

What we did
Research with agents, distributors and customers; a guided qualification flow and a configurator with automatic quotes.

Result
Qualification from six weeks to three days, +60% features sold per order.
Machinery & Manufacturing

A Corporate HMI to unify an industrial group

Challenge
A converting and packaging group with diverse machines and no common interface.

What we did
Field interviews in Italy and China, 21 interaction flows, more than 50 screens and a component library.

Result
A single standard HMI for the whole portfolio and a first step towards a group design system.