How Marsh McLennan Built Its Own AI for 90,000 Employees at 1% of Microsoft's Price — and Saved 8 Hours a Week
When Marsh McLennan's leadership team weighed the cost of licensing Microsoft Copilot for its entire global workforce, the math was stark. Enterprise AI subscriptions at scale run into tens of millions per year. So the global professional services firm did what most companies cannot: it built its own tool — and delivered it at roughly 1% of what the Microsoft alternative would have cost.
The tool is called LenAI. Today it serves 90,000 employees across four business units, processes approximately two million AI requests per month, and has become central to how the company works.
For HR and operations leaders evaluating enterprise AI, LenAI is one of the clearest case studies in the build-versus-buy debate — and the lessons apply regardless of whether your team has the engineering capacity to build from scratch.
The Problem: Enterprise AI Is Expensive and Risky
The business case for generative AI is no longer in question. The obstacle is execution. At enterprise scale, three barriers surface repeatedly:
Cost. Microsoft 365 Copilot licenses start at $30 per user per month. For an organization with 90,000 employees, that translates to over $32 million annually — before customization, training, or integration work.
Privacy. When employees paste client data, financial models, or legal documents into third-party AI tools, that data leaves the building. For a risk management and insurance firm handling sensitive client information, external data exposure is a non-starter.
Rollout velocity. Off-the-shelf enterprise AI products serve every industry. That generality means they rarely match the specific workflows that drive productivity in professional services.
The Solution: LenAI
LenAI is a privately hosted generative AI platform co-developed by Marsh McLennan's Dublin Innovation Center and Oliver Wyman Digital. Built on Microsoft and OpenAI APIs, it runs on Marsh McLennan's own infrastructure — every query, every document, every interaction stays inside the company's security perimeter.
The platform launched in November 2023. Over the following 18 months, the team scaled it from early adopters at Oliver Wyman to the full global organization of 90,000 employees.
LenAI supports approximately 40 production use cases, including:
- Document drafting and summarization — meeting notes, memos, client reports
- Translation — cross-border team communication
- Client research — synthesizing publicly available data into briefings
- Email composition — drafting and refining professional correspondence
- Calculations and data analysis — financial modeling support
- Web browsing — retrieving and summarizing external content within a secure environment
Paul Beswick, Marsh McLennan's Senior Vice President and Global Chief Information Officer, called the tool "a game changer for our teams, enabling them to work smarter for the benefit of our clients."
The Results
The numbers tell a clear story.
Eight hours saved per week. Early adopters at Oliver Wyman reported saving an average of eight hours weekly — a full working day reclaimed for higher-value tasks.
20% reduction in repetitive work. Employees spent one-fifth less time on routine, simple tasks like formatting, summarizing, and data entry.
Two million AI requests per month. Usage at this volume signals genuine adoption, not a novelty pilot. When employees use a tool two million times a month, it has become infrastructure.
AI Academy for all employees. Marsh McLennan established an AI Academy to train its entire workforce on generative AI technologies — ensuring adoption was not limited to technically inclined staff.
The SHRM Executive Network's June 2026 Technology Trends report highlighted the initiative, noting that Marsh McLennan "successfully boosts staff well-being with digital tools, improving productivity and work satisfaction for more than 20,000 employees."
The Strategic Lesson: Build Smart or Buy Smart
Marsh McLennan's story is compelling, but it comes with context. The company has more than 300 AI specialists in its Oliver Wyman Quotient division. It operates a dedicated Dublin Innovation Center. It had the engineering bench, the data infrastructure, and the organizational patience to build, test, and scale a proprietary platform over 18 months.
Most HR teams — and most companies — do not have that bench.
The lesson is not "build your own AI." The lesson is purpose-built AI wins. LenAI works because it was designed for exactly how Marsh McLennan's professionals work: their document types, their compliance requirements, their client workflows. A generic tool would have cost more and delivered less.
The same principle applies to hiring. A general-purpose AI chatbot repurposed for candidate screening will never match a tool built from the ground up for recruiting workflows.
While Marsh McLennan had the engineering capacity to build from scratch, most HR teams don't — and shouldn't need to. For the hiring function specifically, purpose-built AI tools like OVI deliver the same "AI does the work, humans make decisions" advantage without requiring a single line of internal code. OVI starts at $99 per month and keeps humans in the loop on every hiring decision — purpose-built for recruiting the way LenAI is purpose-built for Marsh McLennan.
What is LenAI?
LenAI is a privately hosted generative AI platform built by Marsh McLennan's Dublin Innovation Center and Oliver Wyman Digital. It runs on Microsoft and OpenAI APIs but keeps all data on Marsh McLennan's own infrastructure, eliminating third-party data exposure.
How much does LenAI cost compared to Microsoft Copilot?
Marsh McLennan reports that LenAI operates at approximately 1% of what enterprise Microsoft licensing would cost for the same workforce. For a 90,000-employee organization, Microsoft 365 Copilot licenses alone would exceed $32 million per year.
What results has LenAI produced?
Early adopters at Oliver Wyman save an average of eight hours per week. Across the organization, employees spend 20% less time on repetitive tasks. The platform processes approximately two million AI requests per month.
How can other companies replicate this approach?
Most organizations lack the 300-plus AI specialists and dedicated innovation center that Marsh McLennan used. The transferable principle is to choose purpose-built tools that match specific workflows rather than licensing generic enterprise AI. For hiring, that means selecting tools designed for recruiting rather than adapting a general productivity AI.
What does this mean for HR AI strategy?
LenAI demonstrates that the highest-ROI AI investments are purpose-built for specific functions. HR leaders should evaluate whether their AI tools were designed for HR workflows or are generic platforms with an HR module added on.