From isolated AI pilots to AI truly integrated into the business
Most of the companies I work with started with a generative AI pilot or a simple automation case. But what we are seeing now is a change of scale:
- At a retailer I work with, AI has gone from being a trial in customer service to being part of procurement, dynamic pricing and staff planning.
- In banking, we have seen how models that started as simple back-office aids have made the leap to regulatory and operational processes.
- The pattern is clear: AI is no longer ‘the novelty’ but has become ‘the invisible infrastructure’.
The cloud, data and the edge are becoming the basic operational fabric
In several industrial and smart city projects, I have seen how the combination of cloud and edge computing and IoT is no longer up for debate: it is taken for granted.
- In an urban mobility project, for example, it was not feasible to wait for the cloud to process video: the logic had to occur at the edge due to latency and cost.
- In banking, on the other hand, the trend is towards hybrid environments where highly regulated on-premise data coexists with AI and analytics capabilities in the cloud.
Less focus on ‘adopting technology’ and more on governing data and empowering people
In virtually all the clients where we have provided strategic support, the biggest limitation to transformation is not technology: it is the organisation.
There is a lack of data governance, clear roles, accountability frameworks and digital skills. That is why we are seeing increasing interest in CDOs, data offices, governance models and adoption programmes.
In short, it is the real industrialisation of AI, data and the cloud within processes that generates tangible impact.
How does technological innovation affect the way we work in this field?
Technological innovation is radically changing the way both our clients’ internal teams and our consulting teams work. What I see every day:
‘Augmented work’: teams perform better, not because they work more, but because they work better. In many projects, we have seen how generative AI enables analysts, salespeople and operations managers to produce higher quality work in less time:
- At an insurance company we work with, claim analysis time was reduced by more than 40% thanks to intelligent summaries.
- In banking, regulatory documentation went from days to hours with support tools.
- AI does not replace, but it does change the type of value that each person brings.
Decisions are based more on data and less on personal intuition: I have seen this especially in highly operational sectors: transport, retail, and energy.
Managers continue to contribute judgement and experience, but now with dashboards, predictive models, and simulations that allow them to anticipate scenarios, not just react to them.
More hybrid and cross-functional teams: The projects that work best—and I speak from direct experience—are those where, for example, business, technology, data, security and people work together from day one.
It is becoming increasingly difficult to separate ‘technology’ from ‘operations’.
More distributed work, but more results-oriented: With many clients, we see how collaborative tools enable more agile processes, but also generate greater pressure to justify the impact of the work.
What role does collaboration between different players in the sector play in achieving better results?
Collaboration between different players is not just a nice concept: it is key to making projects work. I say this because I see it in the field:
The problems are too complex to tackle alone: In a recent smart city project, for example, the following participated: the city council, us in various roles (data, IT, communications), manufacturers and a mobility company. If one fails, the whole project falls apart. The interdependence is total.
Ecosystems accelerate innovation
For a client in the healthcare sector, collaboration between the hospital, a computer vision start-up and a cloud provider made it possible to automate part of the medical image analysis in weeks, not months. No single player could have achieved this so quickly on their own.
Collaboration allows risks to be shared and common standards to be defined
I see this in industrial and logistics projects. Sometimes it is not so much that each company adopts a technology, but that they all adopt the same technology in order to interoperate. That requires sitting down together, not competing.
Public administration is beginning to play a structural role: especially in data (with the famous data spaces), AI and security. I have seen projects blocked until the regulatory framework is clarified, and others unblocked when the regulator becomes a catalyst rather than a brake.
Advice for organisations that want to adapt to new challenges: The advice I usually give is based on real experiences, on projects that have gone well… and on others that have learned from their mistakes:
Start with a real problem, not a technology: In a large industrial company, for example, we ruled out using AI in maintenance because the ROI was insufficient… and instead discovered huge savings by automating the back office.
The difference was focusing on the real pain, not the trend.
Pilot quickly, but with the intention of scaling up: What works is what can be industrialised, not the pretty demo. At a banking client, the pilots we defined with clear metrics went into production within months; the vague ones died.
Invest in data and governance before promising ‘magic’: I have seen AI projects fail for one very simple reason: the data was not ready. Without quality, without governance, without integration… AI does not fix anything.
Accompany people from the beginning: In all the projects where we have done change management —training, communication, adoption measurement— the technology was adopted better and faster.
When this is not done, internal resistance becomes the biggest obstacle.
Build internal capabilities, don’t always rely on the supplier: We have developed internal data and AI teams for several clients that are now autonomous thanks to a training and change management programme. This gives them a competitive advantage that cannot be replicated by outsourcing alone.
Accept that digital transformation is never-ending: It is not a ‘phase.’ It is an ongoing capability. The most mature organisations are those that have understood that adapting is not a project, it is a way of working.
