97%
Copilot adoption rate across the business
Case study
SoftwareOne turns Customer Zero lessons into a path to AI value
SoftwareOne became its own first AI customer, learning how to turn Copilot adoption into measurable results before applying the lessons to customer projects.
Rolling out AI licenses is the easy part. The harder work is changing how people work, keeping cost and governance under control, and connecting AI to the systems where work actually happens. These are the barriers that hold many businesses back on their AI journeys. As one of the first companies worldwide to earn Microsoft’s Frontier Partner Badge, SoftwareOne decided to test its own advice first. As Customer Zero for Microsoft’s AI capabilities, it moved AI from experimentation into daily production, reaching a Copilot adoption rate of 97 percent. This case study follows that journey, detours included, and the six lessons SoftwareOne now uses to help customers identify the right starting point, prove value safely and scale what works.
80%
Measured reduction in effort for contract-data extraction in IT asset management
4.8 hours
Minimum monthly time-saving per user measured in M365 apps
- Client
- SoftwareOne
- Industry
- Information technology
- Platform
- Microsoft Azure
- Services
- Data and AI Digital Workplace Services: Services for Microsoft 365 Copilot
From 50 Copilot pioneers to 97 percent adoption at SoftwareOne
SoftwareOne was one of first companies worldwide to earn Microsoft’s Frontier Partner Badge, which recognizes organizations that place AI at the center of their operations and deliver AI-first solutions.
“Naturally, we speak with customers about AI every day,” says David Tannheimer, Head of Internal AI Advisory at SoftwareOne. “That creates a responsibility: before telling others how to transform, we must prove that we can transform ourselves.”
As Customer Zero for Microsoft’s AI capabilities, SoftwareOne tackled barriers that hold many businesses back on their AI journeys. This case study follows its progress from AI adoption to measurable business value and shows how those lessons now inform its support for customers’ AI projects.
“This wasn’t a linear journey,” Tannheimer says. “At times, we created our own friction, especially through tool sprawl. Our progress came not from getting everything right, but from learning quickly when we did not.”
The Customer Zero journey started in October 2023 with just 50 Copilot pioneers; in November 2026, SoftwareOne counted 6,522 Copilot licenses with 6,342 active users, giving an adoption rate of 97 percent.
“That did not happen simply because we made licenses available,” Tannheimer points out.
SoftwareOne expanded its Copilot deployment in controlled waves, starting with sales and moving on to consulting and delivery. Throughout, it monitored what worked and what didn’t, and a monthly license-recycling program pulled back licenses that were sitting idle.
“Customer Zero is an operating discipline,” says Tannheimer, “and because we were disciplined about reclaiming unused licenses, we now sit at that 97 percent adoption rate rather than carrying a large number of dormant ones.”
To reinforce adoption, SoftwareOne created a 30-day license trial model, as well as regular Copilot drop-in calls and inspiration sessions for new users.
Employees need relevant support, consistency and time to build confidence, SoftwareOne learned.
“Even an excellent AI solution fails if people do not understand it, accept it and build it into the way they work,” Tannheimer says.
Choose the right level of complexity, and put AI where work happens
As demand grew, SoftwareOne established four lanes for AI use and development. The rule is simple: start with the lowest lane that meets the need and move up only when necessary. Lane 1 is open to everyone, with no license or approval required. Lane 2 supports licensed self-service development. Lane 3 covers business-owned automation built with community experts or SoftwareOne. Lane 4 is reserved for funded, governed, business-critical solutions developed with IT. “Each step increases effort, data access and governance,” Tannheimer emphasizes, “and most needs never reach lane 4.”
Employees can also submit ideas and business cases through an AI funnel. Related proposals are grouped around shared business needs, reducing duplication and directing investment toward common solutions. As of November 2026, SoftwareOne had built in excess of 12,000 agents, predominantly in Copilot Studio and Agent Builder, but also for SharePoint and third-party apps.
A shared lifecycle governs every AI tool and use case, making adoption more consistent and cost-effective.
“The final step – RETIRE – is the one most organizations skip,” Tannheimer comments, “and that is how the AI jungle grows.”
The value did not come from the model alone. It came from placing the result directly into the operational system, where the work actually happens.
Head of Internal AI Advisory, SoftwareOne
The clearest operational lessons emerged when SoftwareOne connected AI directly to existing systems. Its Delivery Team previously read incoming emails, identified the case subtype, account number and manufacturer, and manually created a case. AI now extracts that information and creates the case directly in Delivery CRM, cutting the average email-to-case time from hours to seconds.
“The value did not come from the model alone,” Tannheimer says. “It came from placing the result directly into the operational system, where the work actually happens.”
The first win of the Customer Zero experience was a minimum monthly time-saving of 4.8 hours per user measured in Microsoft 365 apps.
Another proven outcome: the effort to extract contract data for IT asset management was cut by 80 percent.
AI-driven pricing – rather than inconsistent manual pricing – has modeled potential to increase net revenue by 2% to 2.5%, while a license renewal initiative projects a drop in revenue leakage of up to 2 percent.
SoftwareOne achieved these results over time, learning some hard lessons along the way.
The lessons to pass on to customers
SoftwareOne identified six broad lessons it would share with its past self.
- AI is not a tool rollout
Deployment provides access, but value depends on changing how people work. SoftwareOne learned to sustain adoption through role-specific examples, community support and renewed guidance as capabilities evolved. - Not every problem needs generative AI
The solution must fit the business need, data, risk and economics. A reusable prompt or existing pattern can sometimes deliver the required outcome without a complex custom build.
- Data readiness decides trust
Reliable outputs depend on connected enterprise data, appropriate context and reusable instructions. SoftwareOne revisited its foundations when data became a constraint, learning to address those weaknesses before expanding a solution.
- Governance must enable progress
Guardrails should support safe experimentation. SoftwareOne involved legal, data protection and security colleagues in assessing tools, with clear ownership and a lifecycle covering evaluation, testing, scaling, monitoring and retirement.
- Cost must be designed in
Licenses, tokens, model choices and support effort require active financial management from the outset. Overlapping tools demonstrated how quickly support, integration and procurement demands could bloat costs. - Consistency beats fear of missing out
Changing direction with every new model can undermine confidence and disrupt working habits. A stable direction gives teams time to build capability and integrate AI, while allowing course corrections when evidence warrants them.
AI success is not about finding the best model. It is about changing how people work and proving value one business outcome at a time.
Head of Internal AI Advisory, SoftwareOne
Customers face many of the same challenges; SoftwareOne has already worked through them.
Its frontier ambitions now extend to agents with their own identity and permission, with accountability firmly in human hands at every step.
“We advance with agentic AI using the same principle that has served us well so far in our Customer Zero journey. AI success is not about finding the best model. It is about changing how people work and proving value one business outcome at a time.”
“We now use our Customer Zero cycle to help customers put all this into practice,” Tannheimer concludes. “Identify the right starting point, prove value safely, and scale what works.”
Our progress came not from getting everything right, but from learning quickly when we did not.
Head of Internal AI Advisory, SoftwareOne
Ready to turn AI ambition into measurable value?
Whether you are rolling out Microsoft 365 Copilot or planning your first AI agents, our experts can help you identify the right starting point, prove value safely and scale what works.
Ready to turn AI ambition into measurable value?
Whether you are rolling out Microsoft 365 Copilot or planning your first AI agents, our experts can help you identify the right starting point, prove value safely and scale what works.