20,000 Copilot licenses. 20,000 people trained. Adoption stays flat, so leadership approves another round of training.

The dashboard shows usage, but it says nothing about whether anyone actually wants to open the tool, or whether your client can now fact-check your work faster than you can justify your fee.

Bora Ger, Global Lead Human AI Advantage at Capgemini Invent, has watched this cycle repeat across organizations enough times to name it: "You really trained for tolerance, and not for adoption."

Capgemini Invent, a 17,000-person consultancy inside a wider 340,000-person group, replaced that measurement with four metrics borrowed from how NASA designs interfaces for astronauts.

TL;DR

• AI upskilling budgets are getting cut while skills gaps widen, because most training builds tolerance, not adoption.
• Capgemini Invent measures cognitive load, stress, trust, and performance instead of course completions, using an index modeled on NASA's astronaut research.
• An internal "non-hackathon" put no-code agent tools in 800 new employees' hands and produced more than 14,000 working agents in weeks.
• Consulting pricing is moving from time-and-materials toward fixed-price and outcome-based models built on hybrid human-AI staffing.
• The 12-month roadmap for a boutique firm starts with finding your internal champions, not writing a strategy deck.

You're Training Your People to Tolerate AI, Not Adopt It

You roll out 20,000 Copilot licenses. You train 20,000 people. Adoption stays flat, so you train more people. Bora Ger has watched this cycle play out countless times.

Bora Ger is Global Lead Human AI Advantage at Capgemini Invent, the roughly 17,000-person strategy and transformation arm of the wider Capgemini Group, a workforce of more than 340,000 people. His job is not adjacent to AI. It is the design of how humans and AI work together at a scale most firms never reach.

Here's what tolerance looks like from the inside, in his words: "Jesus, this is really annoying. This does not really deliver on what I expected. This can do better because I'm looking at other solutions privately." People keep using the tool because they have to. They don't reach for it because they want to. That distinction is invisible on a training completion dashboard, and it's the entire problem.

Your clients are also catching up faster than your training budget is. Bora watched this happen in real time in the market he came from, academic research and expert-insight services, years before consulting felt it. "There is no way if you have a smart person sitting in front of you, that you can come in with a half-assed report, or even a sophisticated one, that is too generic, not thought through enough." Your client can now do that themselves, if they've spent the time.

"You cannot come in and be like, 'Oh yeah, I know more than you do because I'm ahead of the curve,'" he says.

Capgemini Invent built a way to measure that gap, between people using AI and people getting value from it, directly.

What Capgemini Invent Measures Instead of Training Completion

Cognitive load. Stress. Trust. Performance. Those four words sit at the center of how Capgemini Invent now judges whether an AI solution is actually working, and none of them appear on a typical training dashboard.

Bora walks the distinction through three stages. AI literacy means understanding the rules: what the technology is, how to stay compliant, how not to break the company. Necessary, he says, but not sufficient. AI fluency means competence inside your specific role, using a tool well enough that you're somewhat better at your job. Still limited. The stage Capgemini prioritizes is what it calls human-AI chemistry, the quality of the ongoing interaction itself. "Chemistry can mean good chemistry, toxic, explosive. We need to be very deliberate about that."

The four pillars behind the Human-AI Chemistry Index came from an unlikely place: NASA. Bora studied how the space agency designs interfaces for astronauts, environments where a wasted second logging in or clicking away a cookie is not a UX annoyance, it's a mission risk. "The astronaut's focus needs to be on the mission... That's a design question. That's a performance question, which then lets the astronaut trust the system in space." He adapted the same logic to enterprise AI:

This framework is part of Capgemini's Resonance AI Framework, launched company-wide in 2025, in which "human-AI chemistry" is one of three named strategic pillars the entire firm now organizes AI transformation work around.

Run this against your own firm before you read further:

• Does using this tool give your team more capacity for judgment work, or less?
• Does anyone on your team avoid a tool you rolled out, and do you know why?
• Would your team trust this tool's output enough to send it to a client unreviewed?
• When did you last measure this tool's actual response time or error rate, rather than its usage count?

If you can't answer one of these for a tool you've already paid for, the gap is in chemistry, and no amount of additional training will close it.

800 People Built Their Own AI Tools in 3 Weeks

Capgemini built an internal, no-code "agentic workbench," a platform that lets any employee, not just engineers, connect to real company data (SharePoint, Outlook, Microsoft Graph, the open web) and build working agents in plain language. To activate it, the firm ran what it called a "non-hackathon": 3.5 weeks, open to anyone regardless of technical background, no coding required.

The results, in Bora's words: "I know that we had around about 800 new users on their agentic workbench, which now has, I think, something about 3,000 users, more than 14,000 agents built already on the workbench."

How Capgemini got there explains more than the total itself. A finance or marketing person with zero AI background didn't need to file a ticket with IT to test an idea. "You don't have to ask somebody from the IT department or from the AI department... you can go on that platform, build something based on your expertise without the technical expert next to you." That's the cognitive-load and trust pillars showing up as behavior, not survey results.

The same pattern shows up in client delivery. Bora described engagements, in data exploration and anomaly detection specifically, where autonomous agents now do the bulk of the work traditionally staffed by an offshore team, with one or two consultants managing the client relationship and reviewing output. His own framing of the commercial logic: a client facing a traditional proposal against one "for a tenth or a fifth of the price" that's faster and still reliable has an easy decision to make, even if the traditional approach feels safer on paper.

Here's the backdrop those numbers sit against right now. McKinsey's CEO, Bob Sternfels, told CES in January that the firm runs roughly 25,000 AI agents alongside 60,000 employees. Bora says: "That might be even true, but I'm not so much interested in the vanity metrics of how many agents you have deployed. I'm interested in what the results are."

Meanwhile Salesforce, a company built on selling enterprise software at scale, has shipped three different pricing models for its Agentforce product in about 18 months, a sign that even the most resourced AI vendors in the market haven't settled on how agentic work should be measured or sold. Bora pointed to Salesforce unprompted as an example of a market leader still finding its footing. For a reader benchmarking their own AI investment, the number to watch isn't the agent count on either side, it's that even the best-funded players in this market are still guessing at how to price and measure agentic work, which means a competitor's headline figure is not evidence that their approach works.

The Commercial Model You're Still Pricing Like It's 2019

Time and materials is not a safe default anymore. He says: "Traditional commercial models like times and material, that's out of the window. That happens here and there maybe still, but now definitely fixed price, definitely outcome-based, definitely value-based." Once part of delivery is done by agents rather than staffed hours, billing by the hour stops making sense for either side of the table.

The same logic applies to your own technology stack. Bora's team runs a constant "make or buy" conversation rather than a one-time decision, because the tools worth building in-house versus buying keep changing underneath you. "We can't be like, 'Oh, we have this frozen status. And now we unfreeze and we go to the next frozen status.' That's not something we can do in the agentic age."

2 questions worth pondering before your next proposal:

1. When did you last renegotiate a contract's pricing structure, rather than just its day rate?
2. Could your firm adopt a new AI tool this month without a six-month procurement or approval cycle?

If the honest answer to either is "not recently" or "no," you're pricing and building for a market that no longer exists.

Find Your People, Not Your Strategy: The 4-Move Playbook for the Next 12 Months

What should a managing director of a boutique or mid-size consulting firm actually do to be in a meaningfully better position 12 months from now?

Bora's answer was 4 concrete moves, given roughly in order.

1. Find your people first. Every firm already has a small group experimenting with AI on their own time, even without a mandate. "Where are people? If you are ready to put the chips on the table, who will support you? Who are the first adopters?" Identify them before you design anything.
2. Fast-track access, don't run a year-long approval process. Bora says: "It has been done already by companies far more regulated than you are. In the finance industry, in the healthcare industry." Give your champions tools now, in a controlled environment, rather than waiting on a full-year procurement cycle. If they can move fast, so can you.
3. Build controlled momentum, not chaos. This is not permission to break things. "This is about controlled exploration. With control I mean don't do stupid things... not try everything and break everything and lose trust with the client." Once you have a critical mass of people building real solutions, behavior leads and culture follows.
4. Set strategy and lean governance once momentum exists, not before. Governance built too early calcifies around assumptions that won't survive contact with the technology. Bora's framing: a living system, revisited constantly, not "static for the next five years."

The Bigger Picture

Bora doesn't call himself an optimist or a pessimist about consulting's future. He calls himself a realist: "Not all consulting companies will survive. That's the reality of things," he says.

Read plainly, that's an invitation to start now, while the outcome is still yours to decide. If you haven't started, there's still time, but standing still is the one position that reliably gets worse.

You already know where your own experiments could start. The only real question is whether you'll let them start before someone forces the decision for you.

Frequently Asked Questions

◾️ What is Capgemini's Human-AI Chemistry Index?

It's an internal measurement framework Capgemini Invent uses to judge whether an AI tool is actually working, based on four pillars: cognitive load, stress levels, trust, and technical performance. It was inspired by NASA's research into astronaut mental workload and is part of Capgemini's company-wide Resonance AI Framework, launched in 2025.

◾️ What is the difference between AI literacy, AI fluency, and human-AI chemistry?

AI literacy means understanding what a tool is and how to use it compliantly. AI fluency means being competent enough with a tool to perform slightly better in your existing role. Human-AI chemistry, the stage Capgemini focuses on, is the deliberate design of the ongoing interaction between people and AI so it reduces stress and cognitive load rather than adding to them.

◾️ Why do AI training programs often fail to increase adoption?

Traditional training measures completion and usage, not whether people actually want to use the tool. Capgemini Invent's Global Lead for Human AI Advantage, Bora Ger, describes this outcome as training people for "tolerance," continued use out of necessity, rather than genuine adoption driven by real value.

◾️ What is Capgemini's Resonance AI Framework?

Capgemini's company-wide strategic framework for enterprise AI transformation, launched in 2025, organized around three pillars including "human-AI chemistry," the deliberate design of human-AI interaction across an organization's workflows and culture.

◾️ How many AI agents has Capgemini's agentic workbench produced?

According to Bora Ger, an internal three-and-a-half-week initiative called a "non-hackathon" brought roughly 800 new users onto Capgemini's no-code agentic workbench, which now has around 3,000 total users and more than 14,000 agents built.

◾️ Is time-and-materials pricing still viable in consulting?

According to Bora Ger, traditional time-and-materials pricing is being replaced across the industry by fixed-price, outcome-based, and value-based models built on hybrid human-AI staffing, as AI changes how much staffed time an engagement actually requires.

◾️ How should a small consulting firm start using AI without a large budget?

Bora Ger's advice is to identify employees already experimenting with AI on their own initiative, give them fast-tracked access to tools without a lengthy approval process, and build momentum through controlled exploration before formalizing governance.

◾️Will all consulting firms survive the shift to AI?

Bora Ger, Capgemini Invent's Global Lead for Human AI Advantage, describes himself as a realist rather than an optimist or pessimist on this question, stating plainly that not all consulting firms will survive the current transition.

About the guest:

Bora Ger is Global Lead Human AI Advantage at Capgemini Invent, the roughly 17,000-person strategy, technology, and design consulting arm of the Capgemini Group. He leads the design of human-AI interaction models across Capgemini's internal transformation and client delivery work, including the Human-AI Chemistry Index discussed in this article. Connect with him on LinkedIn.

About Capgemini Invent:

The strategy, technology, data science, and creative design consulting business of the Capgemini Group, headquartered in Paris. Capgemini Invent operates across more than 60 locations worldwide and is part of a wider Capgemini Group workforce of more than 340,000 people.

This article is based on an episode of the LEADERS IN CONSULTING Podcast, hosted by Sammy Gebele, Founder of SAWOO.

If you want to hear the full conversation behind this analysis with Bora, you can find the episode in the podcast section.

If you’d like to stay in the loop on future deep dives and community updates, you can subscribe to the newsletter here.

For more perspectives on how leaders are navigating challenges like this, follow LEADERS IN CONSULTING on LinkedIn.