According to McKinsey's latest research, professional services leads all sectors in generative AI adoption—up from 33% in 2023 to 71% in 2024. AI consulting is expected to account for 40% of revenue by 2026.
The narrative writes itself: AI will transform how services firms operate. Junior work gets automated. Insights get generated at scale. The firms that embrace AI win; the laggards lose.
There's just one problem: it's not working out that way.
The Messy Reality
Talk to services leaders implementing AI and you'll hear a different story. Yes, tools like ChatGPT and Copilot are everywhere. Yes, analysts are generating first drafts faster. But the operational problems persist.
Why? Because AI amplifies whatever system it's deployed into. If your service definitions are vague, AI generates vague outputs faster. If your pricing is arbitrary, AI makes arbitrary pricing at scale. If delivery knowledge lives in people's heads, AI can't access it.
"You can't automate chaos. You can only automate structure—and most services firms don't have any."
The Structure Problem
Most professional services firms have never formalized the basics:
- What exactly do we sell? — Not the marketing description. The actual service components, inputs, outputs, and dependencies.
- How do we price it? — Not "it depends on the deal." The logic that connects scope to price to margin.
- What did we learn? — Not individual memories. Captured patterns from past engagements that inform future ones.
Without these foundations, AI has nothing solid to build on. It's reading unstructured text, generating plausible-sounding outputs, and hoping humans catch the mistakes.
The Obelisk Is Coming
Harvard Business Review recently described how AI is creating a "leaner consulting model"—the obelisk. Junior analyst work gets automated. The pyramid flattens. Fewer people do more work.
This is real, and it's happening. But here's what the articles miss: the firms that successfully make this transition aren't just adopting AI tools. They're restructuring how work flows through the organization.
When you have:
- Encoded service definitions that AI can read and reason about
- Structured data from past engagements for AI to learn from
- Clear handoffs between stages where AI can assist vs. where humans decide
Then AI becomes genuinely useful. Not as a replacement for human judgment, but as a force multiplier for it.
Human-First AI
There's a reason we call our approach "human-first AI." It's not marketing. It's a design principle.
AI in Servantium operates under constraint. It reads structured operational data, not free-form text. It surfaces patterns that exist in your encoded practice. It recommends actions grounded in your firm's actual history.
No hallucinated scope. No invented context. No synthetic confidence.
The goal isn't to replace the expert judgment that makes services valuable. It's to give that judgment better inputs, faster.
The Right Order of Operations
Here's what we've learned building Servantium and working with services firms:
1. Structure First
Before you can use AI effectively, you need to encode what you actually do. Service definitions. Pricing logic. Delivery patterns. This isn't AI work—it's the unglamorous work of building operational foundations.
2. Data Second
Once you have structure, you can start capturing data that means something. Actuals vs. estimates. Delivery outcomes. Client feedback. This becomes the learning loop that makes AI useful.
3. AI Third
With structure and data in place, AI can actually help. Pattern recognition. Anomaly detection. Intelligent suggestions. But it's earning its place based on your reality, not generating plausible fiction.
Structure before automation
See how Servantium builds the operational foundation that makes AI genuinely useful.
The Opportunity
The services firms that will win in the next decade aren't the ones that adopt AI fastest. They're the ones that build the operational infrastructure AI needs to be useful.
Right now, that infrastructure barely exists. Most firms are trying to layer AI on top of spreadsheets and tribal knowledge. It's not working.
The firms that invest in structure now—while competitors chase AI hype—will have a compounding advantage. Their AI will actually work because it has something real to work with.
AI is a suit of armor for services teams, not a replacement for them. But armor only works if you build the body underneath.
