Pietro Piga

Pietro Piga

Global Salesforce Platform Lead & AI Product Owner
Multinational industrial group, energy division

I have no incentive to make you buy.

In short

Pietro Piga. I review AI, CRM, data and automation proposals for the people who have to buy them - Salesforce and Agentforce included, but not only those. I run the global Salesforce platform of a multinational industrial group, with a $1M project budget. I do not resell licences and I do not deliver implementations: I have no incentive to make you buy.

Do you have a proposal on the table? Thirty minutes to work out what is missing.

Book thirty minutes

No cost, no commitment, nothing to prepare.

Scope

~$5bn

of annual revenue carried by 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, with an agent built in-house

Technologies I work with

  • Salesforce
  • OpenAI · ChatGPT
  • Claude
  • Gemini
  • Perplexity
  • SharePoint
  • Relevance AI · Lovable · Emergent · Custom GPTs · MCP

These are the tools I use to assess a proposal, design an architecture and build a prototype. It is not a list of things I install for you: execution stays with your team or with a partner.

From evaluation to architecture

AI Strategy & Governance
AI roadmap, operating model, security and approval criteria, adoption programme. Who decides what, and who holds the authority to say no.
CRM & Business Architecture
CRM architecture, large-scale modernisation, steering committees and architecture review boards, realised-value measurement. Salesforce, Agentforce and Data 360 live inside this.
AI Agents & Automation
Agents on Agentforce, agents on Microsoft 365 and SharePoint, Gemini Gems, automation and MCP integrations. Designed, evaluated and instructed - not delivered by me.
Knowledge Systems
Queryable knowledge bases, intelligent search, AI that answers from your own documentation rather than from the web.

A system integrator sells you the implementation. I have bought what they are selling you.

You have a proposal on the table: an AI project, a data platform, a CRM extension, an automation programme. It is well written, the numbers look reasonable, and there is nobody in the company who can tell you whether it holds up. You have to decide within weeks. Asking the vendor whether their own offer stands is not a check. Asking a second vendor gets you a second proposal, not an opinion. And nobody wants to be the one who says "I did not follow that" in the meeting.

I do your job, not theirs. I run the Salesforce platform of a multinational industrial group, I decide how the budget is spent and then I go and explain to a steering committee where the money went. I buy the same things they are selling you, from the same vendors.

Three things a system integrator cannot give you.

I know where projects stall
I have watched them stall from the inside. Not during the build: afterwards, when nobody could say any more whose the data was, who was allowed to see what, and how you would show it had been worth anything.
I know what you will be asked after signature
The CFO, security, legal. I am the one putting those questions to vendors, every week. I also know which answers are smoke.
Nobody pays me to make you buy
I do not resell licences. I have no quota to close at quarter end. I have no people on the bench to put on a project. That is why my opinion is worth something.

Background.

NowGlobal Salesforce Platform Lead & AI Product Owner at a multinational industrial group. Team of 9.
2024-2025Lead AI Transformation & Business Architect, same group. AI agent on Teams and SharePoint: internal support tickets down by more than 30%.
2022-2024Business Architecture Specialist, Accenture.
2020-2021Salesforce Project Manager, Seven IT.
Certifications12 Salesforce, including Data Cloud Consultant and AI Associate.

Quoted as a source in Salesforce Ben across three articles on Agentforce. The quotes · 14 things I have built

How I work.

Before committing a budget, you prove the hypothesis holds. Then you pick a number, you write down the names of the people who can say no, and at the end you count.

  1. 01 Validate first, commit second A working prototype costs days. An implementation proposal costs six figures. You test the hypothesis with the first, before paying for the second.
  2. 02 One use case, one number agreed upfront Just one, with the measure chosen before starting and a written criterion for stopping. Large programmes rarely fail: they continue, and absorb budget, because stopping them requires someone to admit a mistake.
  3. 03 Whoever can say no, named at the start Security, compliance, final authority: three names on a page before you begin. If you do not choose them at the outset, you discover them halfway through - when they say no and everything stops.
  4. 04 Value gets counted How many hours a process gives back, and how far the number of people asking for help to use it drops. Those are the two figures a steering committee accepts, because they can be counted. Features delivered is not a measure of value.

What you get, and what it costs.

Prices are published. If you have to negotiate on price, it means we have not understood the problem yet.

PackageWhat you getTimingPrice
Independent review of a Salesforce or AI proposal Three to five pages: what is missing, ordered by what ignoring it costs you, and the questions to put back to the vendor. Half a day of analysis, delivered within 5 working days €800-1,000
AI & CRM Readiness Assessment A verdict on data, processes, architecture, governance, skills and value measurement. Then what to do, in what order, with the estimates. 2 days + document + readout call €2,400-3,000
Architecture & Data Assessment What is actually inside your org, which constraints tie your hands, and a technical specification you can hand identically to every vendor. 3-4 days €4,000-5,500
AI & CRM Architecture Advisory Roadmap, architecture and governance without hiring. The days are agreed at the start of the month and that is what they are. 2-3 days a month, one quarter minimum €2,400-3,600 per month
“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.”
Pietro Piga, quoted in Salesforce Ben. The other quotes

Do you have a proposal on the table? Thirty minutes to work out what is missing.

Book thirty minutes

No cost, no commitment, nothing to prepare.

What I do not do.

I do not implement, I do not configure and I do not resell licences.

I hold a full-time role, so I take on few projects a month and that will stay the case. If you are looking for someone to execute, you need a partner who does that. I help you choose them and hold them to account - not replace them.

The real limit, stated before you find it yourselves: I do not know your company. In half a day I read a proposal and the context you give me, not your twenty years of accumulated layers. If the problem sits deeper than that, I say so and propose the right package rather than the most expensive one.

What happens after you book.

  1. 01 Book thirty minutes Pick a slot in the calendar. Nothing to write beforehand. If you have the proposal to hand, keep it open.
  2. 02 I tell you whether I can help On the same call. If your problem is not one of mine, I say so and that is the end of it. It happens, and it wastes neither of our time.
  3. 03 You get a one-page proposal Within two working days. Package, price, delivery date. No twenty-page document, no negotiation.

The questions I get on calls.

What does Agentforce really cost to implement?

The cost is not the licence price. There are six lines: licences, variable consumption by usage, the data platform, integration, data remediation and your own people's time. Proposals typically show two of them. The other four arrive after signature, and they are paid by the buyer.

How do I evaluate a Salesforce AI proposal?

Ask for three things in writing: what is explicitly out of scope, which assumptions change the estimate if they turn out to be false, and how we will jointly measure that the project succeeded. A useful answer comes in writing and names an owner, an expected output and the moment it will be checked.

Why do Agentforce projects fail?

Rarely because the technology cannot be configured. They fail when nobody owns the outcome, when the data is not reliable for that specific use case, when access rights are decided too late, or when value was never measured before starting. Gartner attributes project abandonment to the same four causes: data quality, inadequate risk controls, escalating costs and unclear business value.

Is Data 360 worth it for a mid-sized company?

It depends on where the data that answers your question lives. If it all sits inside Salesforce, Data 360 is not a prerequisite for the first use case. If there is more than one source and customer identity is unresolved, then yes: that is precisely the problem it solves.

What should I ask a Salesforce partner before signing?

The useful questions are the ones that make the proposal verifiable: what is explicitly excluded, which assumptions change the estimate if they prove false, which architecture you chose and why, what it costs over 24 months. And one almost nobody asks: what stays inside our org when the project ends, because whoever owns the platform inherits that maintenance.

Do we need an independent adviser if we already have a Salesforce partner?

The partner sells you the implementation: that is their trade and they do it competently. But once the sum at stake runs into tens of thousands, it is worth having someone alongside you who has no licences to resell and no people to keep busy on a project.

What does an independent proposal review cost?

€800 to €1,000, half a day of work, delivered within five working days. You receive a three-to-five page document with the findings ordered by risk and the questions to put back to the vendor.

Do you also do the implementation?

No, never. I do not implement, I do not configure and I do not resell licences. I hold a full-time role and take on few projects a month. If you need someone to execute, I help you choose them, write the specification and check the work.

Can you have an AI architect without hiring one?

Yes, on a counted number of days and with an end date: two or three days a month for a quarter, then it renews or it closes. It covers roadmap, architecture and governance in the months when they are genuinely needed. It is not continuous presence: my capacity limit is structural and I state it upfront, not afterwards.