Across enterprise organizations, workers are quietly turning to consumer-grade AI tools to fill gaps in their workflows. They’re summarizing documents in ChatGPT, drafting emails in Claude, and running data through platforms your security team has never approved. It’s not malicious, it’s practical. But it’s still a problem for IT leaders.
The fix isn’t to clamp down on this behavior, it’s to channel it. Microsoft 365 Copilot gives enterprises an AI assistant that sits inside the tools people already use, with the governance and security controls you need. But getting from pilot to meaningful adoption? That’s where most organizations stall.
We recently brought together Jane Koh from Microsoft and Kim England from Pearson to share what’s actually working when it comes to driving Copilot adoption at scale. Pearson’s numbers speak for themselves: 3,000 training signups in the first month and over 5,000 employees now holding a Generative AI certification.
Here are five lessons from the conversation that IT leaders and digital workplace teams can act on now.
1. Serve appetizers before the main course
Many organizations get stuck in pilot purgatory. They’re waiting for a transformative ROI case before scaling, and in the meantime, momentum dies and early adopters lose interest.
Jane Koh offered a useful analogy: if you’ve got hungry guests, you can’t make them wait hours for a complex main course. You need to serve appetizers.
In practice, this means deploying out-of-the-box Copilot capabilities and low-code agents that deliver value in weeks, not months. These quick wins keep stakeholders engaged and buy your team the time needed to build more complex, high-ROI use cases.
Pearson applied this thinking to their training rollout. Instead of a single, generic Microsoft 365 training program, they built job-specific sessions. Marketing teams learned Copilot alongside other marketing professionals. Engineers trained with engineers. Every session focused on the tools and tasks people actually use day to day, not theoretical demonstrations.
The approach garnered 3,000 signups in the first month. That kind of momentum is hard to manufacture if you’re asking everyone to wait for a grand unveiling.
2. If you don’t provide AI tools, your employees will find their own
Jane Koh used a term that should make every IT leader pay attention: Shadow AI. It’s the AI equivalent of Shadow IT, and it’s already happening in most enterprises.
When employees see the potential of AI but don’t have access to approved tools, they improvise. They paste confidential documents into consumer chatbots. They use free transcription tools for sensitive meetings. They create workarounds that sit completely outside your security perimeter.
The fix isn’t a policy memo that bans unapproved tools (although having a clear policy is key!). It’s giving people something better within the flow of work they’re already in. Copilot embedded inside Microsoft 365, surfacing answers from your own organizational data through a governed framework, removes the incentive to go elsewhere.
This is where the Unily and Microsoft Graph integration becomes practical rather than theoretical. The Copilot connector (now generally available) feeds Unily content, things like HR policies, corporate news, and operational guidance, directly into Microsoft Graph. Employees ask a question in Copilot and get answers drawn from both systems, with security permissions intact.
“Graph became super important with AI because it’s the semantic index, the organizing layer of your organizational data. We opened that up to external data sources that have connection points. That includes Unily platform data.” — Jane Koh, Microsoft
The practical takeaway: the more relevant data Copilot can access securely, the less reason employees have to look elsewhere. And the less Shadow AI you have to worry about.
3. Kill the vending machine mindset
One of the biggest barriers to sustained AI adoption is expectation. Employees treat Copilot like a search engine: put in a query, get a perfect answer. When that doesn’t happen in their early tries, they write it off.
AI requires iteration. It’s a reasoning engine, not a retrieval system, and the difference matters. Users who learn to refine their prompts, provide context, and build on initial outputs get dramatically better results than those who expect a finished product from a single input.
This is fundamentally a training and culture challenge, not a technology one. Pearson recognized this and segmented their workforce into three groups:
-
Beginners who needed responsible AI basics and general tool awareness
-
Experienced users ready to integrate AI into their daily workflows
-
Advanced users developing new use cases and pushing boundaries
All users started with responsible use policies (non-negotiable), but from there, content depth matched where people actually were. An engineer received different resources than someone in marketing. This stopped the common problem of overwhelming beginners with advanced content while boring experienced users with basics.
Pearson also made learning tangible. They partnered with Microsoft to create a pilot accreditation program where employees could earn a Generative AI Foundations badge. Over 5,000 employees have now earned theirs. When professional development is attached to AI adoption, it stops feeling like an IT mandate and starts feeling like a career investment.
“The disruption was worth it given the gains in efficiency and capabilities. We have to think of AI in a similar construct. But this is a total shift change we’re asking of our employees.” — Jane Koh, Microsoft
4. Let communities lead the way to innovation
Executive sponsorship does matter. That said, Pearson’s experience shows that the real engine of adoption is peer-to-peer advocacy, not top-down mandates.
They launched an AI Ambassadors program as a community within Unily and over 1,000 employees volunteered. These weren’t just IT staff. They came from across the business: finance, marketing, operations, HR. With every new rollout, ambassadors got early access, surfaced pain points before wider deployment, and helped refine solutions in real time.
The community has since evolved into chapters for marketing communications, advanced AI techniques, and specific business functions. As people advance, they help newcomers without leadership orchestrating every interaction. It’s self-sustaining in a way that top-down training programs rarely are.
Critically, Pearson’s leadership played their part by creating psychological safety around experimentation. They communicated explicitly that this was a learning journey for everyone, including them, and that failed prompts and imperfect outputs were expected and acceptable. They even ran image-generation competitions to build familiarity through play before applying AI to serious business tasks.
This bottom-up and top-down approach met in the middle. Ambassadors drove grassroots enthusiasm. Leadership gave explicit permission to experiment. The result was adoption that felt organic rather than enforced.
5. Make IT and HR the new power couple
AI adoption sits at the intersection of technology implementation and people management. Most organizations still treat these as separate workstreams, with IT handling the technical rollout and HR running the occasional change management comms. That split creates friction.
Pearson took a different approach. Their CHRO and CTO presented together at Davos, calling themselves “The New Power Couple.” It wasn’t just a good soundbite. It reflected a genuine alignment where governance policies (IT’s domain) were developed alongside employee engagement strategies (HR’s domain) so that end users experienced a coherent journey rather than disconnected mandates from different departments.
Jane Koh noted that forward-thinking companies are increasingly merging responsibilities between CIOs and CHROs when it comes to digital experience. The technology needs to work securely. The humans at the center of the transformation need support, training, and a reason to care. Neither function can deliver that alone.
For IT leaders, the practical implication is straightforward: if your AI rollout plan doesn’t have HR as a co-owner from day one, you’re building on shaky ground.
Where to start
Pearson didn’t achieve 5,000+ AI-badged employees by accident. They combined secure technology infrastructure, persona-based training, community-led adoption, and genuine cross-functional partnership between IT and HR. None of those elements work in isolation.
For organizations still in the early stages of their Copilot journey, the pattern is clear: provide approved tools before employees find their own, start with quick wins that build momentum, train people based on where they actually are, empower communities to lead, and bring IT and HR together from the start.
Unily’s integration with Microsoft 365 and Copilot gives enterprise organizations the platform to make this happen securely and at scale. If you’re ready to move beyond the pilot phase, the lessons from Pearson and Microsoft offer a practical starting point.
Take the next step
Ready to see how Unily can help you drive engagement and adoption? Discover how our platform integrates seamlessly with your Microsoft environment to deliver a world-class employee experience.
Go Faster. See Unily in action.
