AI is no longer an abstract technology discussion for services businesses. It is actively reshaping the core economics of how they grow, price, and operate.
At a recent Permira-hosted Services CEO Roundtable on AI, we brought together CEOs from across the fund’s global services portfolio alongside external AI and technology leaders. The objective was explicit: move beyond theory and pilots and focus on what is actually working inside services organisations today — and where competitive advantage is already being created.
The discussion cut across professional, financial, information, and outsourced services. Despite different end markets, the conclusions were strikingly consistent.
AI is not a tool upgrade. It is an operating model reset – and the companies moving fastest are beginning to open meaningful gaps.
What follows are the most important takeaways for services CEOs, grounded in real operating examples, case studies, and first-hand perspectives from Permira’s discussion.
1. AI is collapsing cost and time curves faster than most CEOs expect
One of the strongest themes from the roundtable was how quickly AI economics are shifting — and how dangerous it is to plan based on incremental assumptions.
At Octus, a Permira-backed credit intelligence platform serving investment banks, law firms, and buy-side institutions, the cost of running the same AI models fell by ~95% in a single year. That change alone fundamentally altered the company’s innovation capacity, allowing them to run 20 times more experiments on the same budget.
This collapse in inference costs did not just reduce expense — it transformed speed. Teams could iterate continuously, test new workflows, and deploy improvements without rationing AI usage.
At a global scale, the same dynamic is playing out across software development and internal operations. In some cases, AI-enabled teams moved from 6–9-month delivery cycles to weeks, with teams of just three to four senior engineers, heavily augmented by AI.
CEO implication:
Most planning frameworks dramatically understate how quickly AI-driven cost and speed curves are moving. Services businesses that wait for stability before acting are likely to find that the economics have already shifted beneath them.
2. Output is no longer tightly coupled with headcount
Perhaps the most uncomfortable insight for many services leaders was the growing decoupling between output and people.
Historically, services growth has been constrained by hiring capacity, utilisation, and leverage ratios. AI is breaking that relationship.
At Octus, AI has been embedded directly into core research, content production, and internal operations, allowing the business to scale output materially without linear headcount growth.
“This isn’t about replacing people,” said Kent Collier, Founder and CEO of Octus. “It’s about turning them into superhumans — and reducing human error at the same time.”
Taking this further, another participating portfolio company explicitly models sustained 15–18% revenue growth while reducing workforce size by roughly 25%. If achieved, this combination would effectively double enterprise value through operating leverage.
CEO implication:
AI is severing the historical link between revenue growth and labour growth. For services businesses, this creates a once-in-a-generation opportunity to expand margins, reinvest aggressively, or both — while competitors remain people-constrained.
3. Workflow ownership — not tools — is where advantage is being built
A recurring frustration among participants was how many AI initiatives stall at the “copilot” stage. The roundtable consensus was clear: standalone tools do not create durable advantage.
The biggest gains come from companies that redesign entire workflows end to end, rather than layering AI onto legacy processes.
“Most companies are trying to add AI to broken workflows instead of redesigning the workflow itself,” said co-founder of Korza, a specialist AI advisory firm, Zack Kass. “This is not a tool shift. It’s a systems shift.”
At Cielo, a provider of talent acquisition services, agentic workflows now sit at the centre of their entire recruitment process. Powered by a proprietary GenAI platform, CLO.ai, their recruiters now have an agentic assistant from tailored candidate outreach through to role fulfilment.
“Owning the workflow is more powerful than owning the model,” said Marissa Geist, CEO of Cielo.
Several CEOs also noted that businesses embedding themselves into client workflows, not just improving internal efficiency, are far better positioned to defend pricing and relevance.
CEO implication:
Buying AI tools is table stakes. Redesigning workflows — and owning them — is where margins are protected or compressed.
4. Pricing models are already starting to reset
One of the most candid parts of the discussion focused on pricing. AI is making it possible to deliver work historically billed in hours or days in minutes — and clients are already adjusting expectations.
“Clients will assume AI-enabled delivery as table stakes very quickly,” said David McVeigh, CEO of Axiom. “Not as a premium feature.”
As a result, services companies face growing pressure to move away from time-based billing toward outcome- or value-based pricing. AI accelerates delivery, but unless pricing models evolve, efficiency gains simply flow through to clients.
“This isn’t about doing the same work faster,” stated McVeigh. “It’s about doing fundamentally different work — and charging for that difference.”
CEO implication:
AI efficiency without pricing evolution leads to margin leakage. Firms that rethink packaging, outcomes, and value capture early will reset client expectations in their favour.
5. Adoption must be explicit, leader-led, and accountable
Across both success stories and stalled initiatives, one pattern was clear: AI adoption does not scale through enthusiasm alone.
Where AI is delivering real impact, leadership has made adoption explicit, expected, and measurable. AI usage is embedded directly into performance management.
“Most AI failures are ownership failures, not technology failures,” said another Permira AI adviser.
Employees are explicitly asked how they are using AI and what impact it is having. This clarity accelerates learning and normalises AI as part of everyday work.
Several participants also flagged middle-management resistance as a material execution risk, particularly where incentives remain tied to legacy utilisation or headcount models.
Across the Permira Services portfolio, leadership agreed that accountability sits squarely at the top, and the appointment of Chief AI Officers is becoming more prevalent.
CEO implication:
AI progress depends less on tools and more on leadership clarity. When expectations, ownership, and incentives are explicit, adoption follows.
What the leading services businesses are doing differently
- Across the strongest examples, AI is consistently treated as:
- A general-purpose capability, not a siloed initiative;
- An operating model redesign, not a tool rollout; and
- A board-level priority, with clear ownership and urgency.
Zack Kass summarised: “Access to AI is not the advantage. Speed of adoption is the advantage."
