The industry has a name for what comes next: PS 2.0, the AI Economics era, the intelligent services organization. What much of the conversation doesn't have is a clear answer to where the AI gets its data. That isn't a detail, it's the difference between the thesis working and the thesis being a very good conference session. PS 2.0 runs on delivery data, and in many firms the AI is still reasoning about an export.

The Argument and Its Prerequisite

Every version of the PS 2.0 case depends on AI doing useful work inside service delivery rather than around it: predicting which engagements are trending toward overrun, optimizing staffing against skills and availability, drafting the status update nobody wants to write, and spotting a client relationship deteriorating before it escalates.

Technology & Services Industry Association (TSIA), which coined the AI Economics framing, makes the prerequisite explicit. Its State of Customer Success 2026 research concludes that AI only delivers value when the data foundation is unified, and that the bigger barrier usually isn't messy data, it's access. Customer data stays trapped in silos, with sales in the CRM, support in ticketing, customer success in its own tools, and product owning usage data.

Delivery in many services organizations looks the same. Projects are in a PSA, the customer is in the CRM, hours are in a time tool, tasks are in a project tool, and client communication is in email and a meeting recorder. Each was a reasonable choice at the time, and together they mean the AI gets pointed at an export. An export is a photograph of something that moves, so the model answers confidently about a version of the business that existed when the extract ran.

Why Pilots Impress and Don't Scale

That explains a pattern many services leaders will recognize. The pilot goes well, everyone's impressed, and then it never becomes part of how the business runs. TSIA describes the same arc in customer success: tools adopted quickly, pilots that showed promise, and value that was never fully quantified.

The usual diagnosis is change management, but the cause is often simpler. The output couldn't be acted on without verification, and anything that has to be checked before action is a second opinion rather than a system. A risk signal you have to go and check is really just a prompt to do the work you were already doing.

Data Model First, Intelligence Second

The firms getting real gains did the unglamorous part first, bringing projects, resources, time, billing, and client communication into one governed data model, and only then applying intelligence on top of it.

That order is rarely what happens, because consolidating a data model is a long, unglamorous program and buying an AI tool takes a quarter. But the sequence decides whether the second step produces anything durable. Deloitte's 2026 State of AI in the Enterprise survey of more than 3,200 leaders found that only 40% rate their organizations as highly prepared on data management, a share that fell from the year before, even as AI investment keeps rising.

Where It Runs Is Not a Footnote

A second requirement often gets skipped in PS 2.0 discussions, and it gets more serious the larger the firm. Routing delivery data through a separate AI tool introduces a gap between what the model sees and what's actually in the system of record. For a services business, that data includes client communications, commercial terms, rates, and staffing decisions, so the gap is both an accuracy problem and a governance one. Deloitte's same survey found data privacy and security is the AI risk leaders worry about most, cited by 73%.

Where Cloud Coach Fits

Cloud Coach is built for that order of operations. Projects, resources, time, billing, and the customer record live natively in the same Salesforce org, and the AI runs natively on Agentforce, on those live records, under the permissions the admin already maintains. Project Copilot answers questions about project health from live Salesforce data, AI Meeting Manager turns client conversations into tracked decisions on the project record, and Customer Signal Intelligence flags stakeholder health while there's still room to act. Customers report about a 30% reduction in project overruns, which is the kind of operating change PS 2.0 promises when the AI can see delivery as it currently is.

Every vendor has announced AI. The question worth asking each of them is where it actually runs, and on what data.

How We See It

PS 2.0 isn't a model problem, it's a data location problem dressed up as one. Consolidate first. The intelligence is the easier part, and it only works on top of the less glamorous one.

Frequently Asked Questions Related to Software Tech and Professional Services

• What is PS 2.0?

It's the idea, popularized by TSIA, that professional services organizations are moving into an era where AI does meaningful work inside delivery, from predicting overruns to optimizing staffing, changing how services firms operate and make money.

• Why do AI pilots in professional services fail to scale?

Often because the AI works from exports of data spread across several systems, so its outputs have to be verified before anyone can act on them, which keeps it from becoming part of daily operations.

• What does a professional services firm need before adopting AI in delivery?

A unified delivery data model, with projects, resources, time, billing, and client communication in one governed system, so AI can work from live records rather than snapshots.

Sources Cited

TSIA (Technology & Services Industry Association), Darlene Kelly, "The State of Customer Success 2026: Proving Value in the Age of AI Economics," February 11, 2026. https://www.tsia.com/blog/state-of-customer-success-2026-ai-economics

Deloitte AI Institute, "The State of AI in the Enterprise," 2026 report (survey of 3,235 leaders, August to September 2025). https://www.deloitte.com/global/en/issues/ai/state-of-ai-in-enterprise.html Key findings: https://www.deloitte.com/cz-sk/en/services/consulting/research/the-state-of-ai-in-the-enterprise.html