Pietro Piga, Global Salesforce Platform & AI Lead at a multinational industrial group in the energy sector. I work on CRM and AI in large and small companies, with three methods of mine: Value Intelligence, Safe-to-Learn and Portable Operational Memory.
If the work on a CRM can be done from outside it, the product was never the interface. It was the data, the rules and the permissions.
The CRM is not disappearing: it is losing its monopoly on the place you use it from. And what it holds matters more than before.
Three questions before any AI project
I work on CRM and AI inside a multinational and inside ten-person companies. Three methods came out of that difference, one per question.
- Is it worth doing? Value Intelligence: fund measured value, not features shipped. In the CRM modernisation, the hours given back came to 69,147.
- Does it work, and does it work here? Safe-to-Learn: I test the idea in a demo org separate from the company, before asking for budget and real data.
- What stays when the model changes? Portable Operational Memory: governed data, rules and permissions that the company plugs any model into.
One proof, with the number
The knowledge was already there: processes, releases, removed fields, years of documentation. People could not find it, so they opened a ticket.
We made it queryable inside SharePoint and Teams, where people already worked. Ask the question, get the answer with its source.
- roughly 2,000 users worldwide;
- internal support tickets down more than 30% in the two months after launch.
The knowledge was never missing. The road between the knowledge and the people who needed it was.
A second case, without a number. I taught people to set up a scheduled prompt on Copilot: every morning they get an email, click the link and land on the day’s briefing. No new application, the AI sits where the day was already starting.
Outside working hours I have built fourteen AI agents and tools, at my own expense. Three hours to learn what something really costs, before someone quotes it to me at six figures. The fourteen things.
In your company, which AI use case already has a measured number?
Where I am cited
My prediction is that Agentforce will generate a lot of initial excitement but will burn out quickly for many. 90% of stakeholders, including both customers and consultancies, are unprepared.
- Here's What Salesforce Professionals Really Think About Agentforce, Salesforce Ben
- Salesforce Stock Chosen as Top 2025 Pick By Two Wall Street Firms, Salesforce Ben
- Salesforce Stock Downgraded: Is Agentforce Distracting Salesforce's Core Business?, Salesforce Ben
The numbers
- Scope
- ~€5bn
- of industrial sales pipeline behind the platform I run
- Responsibility
- €1M
- annual project budget, mine to defend
- Efficiency
- 69,147
- hours returned to the organisation by CRM modernisation
- Adoption
- -30%
- internal support tickets on roughly 2,000 users, within two months of launch
Previously Business Architect at Accenture. Twelve Salesforce certifications. The full background is on the profile.
Outside the day job
Now and then I talk with companies outside my working hours: owners, CEOs, directors. It is not a commercial activity. From inside a single company some things cannot be seen, and a method tested in one place is tested badly.