Your customers are too busy to adopt AI properly, so they’re adopting it badly. On 4 November, at Arrow’s Agents of Change event in London, Microsoft’s Global Business Development Director David Bowen mapped where they get stuck, and told Partners like you to stop selling models and start selling intelligence and trust. Every Partner in the room heard the same message. Here’s how to make it your own, stand out from the crowd, and get Microsoft to pay for the content through your FY27 Co-op.
Your customers have already adopted AI. They just haven’t told you.
According to Microsoft’s 2025 Work Trend Index, 53% say they don’t have enough time or energy to keep up with their work. They’re too busy doing the day job to stop and learn new ways of working. So they do what busy people always do, and grab a quick fix.
Eight in ten say their employees are bringing their own AI tools to work: personal ChatGPT accounts, free plug-ins, and company data heading wherever those tools take it. Shadow AI. If you look after SMB tenants, you’ve seen it.
That’s your way in. Shadow AI is a live problem, it carries real risk, and you can start fixing it this quarter.
It also proves a bigger point. For many businesses, AI has already passed the point of no return. David, who is responsible for Microsoft’s strategic partners, told the room about one user who couldn’t imagine getting through the day without Copilot. Nobody hands that back. Your customers will get to AI with or without you. What’s still open is whether they get there through a patchwork of personal tools, or with a Partner they trust.
Sell intelligence and trust, and let the models come and go
The AI market is crowded with models and point solutions, and your customers have heard of most of them. David’s advice was to stay off that battlefield. “Models will come, models will go,” he said.
Microsoft is betting on two things that last. Intelligence is the value already sitting in your customer’s business. Trust is the security, identity and governance that let them use it safely, at scale.
Microsoft splits that intelligence into three layers. Your customers won’t know the names yet, so translate them:
- Work IQ: how their people work, drawn from emails, meetings, chats and documents in Microsoft 365.
- Fabric IQ: their business data, brought together so AI understands the numbers behind the organisation.
- Foundry IQ: the knowledge their agents draw on when they act on the business’s behalf.
That’s the anchor for your messaging. Your customer’s competitive advantage is already in their own data, and your job is to turn it into something AI can use safely.
Trust starts with questions. David suggested these: Is the data secure? Who can access it, and when did they last do so? Does anyone know what’s happening in the model? Each one is a blog or a checklist waiting to be written.
Where your customers get stuck
David cited an estimate of 1.3 billion AI agents in the market within two years. We’d bet on multi-agent control being next year’s hot topic.
But the reality is, your SMB customers are nowhere near it. Microsoft maps AI adoption in five stages, and four come before anyone should be letting agents loose on the business:
- Intent. Who will use AI, and to what gain?
- Proof of concept. What do we tackle first?
- Production AI in the periphery. Live AI, delivering incremental gains at the edges of the business.
- Production AI in core processes. Redesigning how the business works, beyond doing the same things faster.
- Agentic AI systems. Agents and multi-agent systems, with everything they bring: development, governance, security, even decommissioning.
David summed up the usual route: a happy first idea, a round of experiments, then a slog to get the board to sign the business case. Many SMBs are parked right there. Some are further back still, with staff using AI informally and nobody in charge of it.
Wherever customers are stuck, there’s an opportunity for you. So find out where each one sits before you pitch anything.
Five drivers of value: pick your ground
Once you know where a customer is stuck, decide what you can fix. Microsoft’s AI transformation guidance, built from its own internal rollout and its work with partners and customers, names five drivers of AI value:
- Business strategy: the business case, the outcomes and the board sign-off.
- Technology and data strategy: getting data ready for AI, which is where Fabric IQ comes in.
- AI strategy and experience: where AI goes first, and how people use it.
- Organisational and cultural change: skills, adoption and the people side of the rollout.
- AI governance and security: who can access what, and keeping oversight as AI scales.
David turned these into questions you could ask a customer tomorrow. Do they have a six, 12 or 18-month AI strategy? Do they know who’s accessing their data?
Then match stage to driver. A customer stuck at intent needs business strategy before anything else. One running AI at the edges needs governance in place before it reaches the core. Pick one or two stages, and the one or two drivers you’re strongest in. That’s your message.
Start where you’re strongest, partner up for the rest
Nobody covers all five drivers, and you don’t need to. Lead with your strongest, win the first piece of work, and expand from there. Get under the skin of your customer’s business, then help them work out where AI could add value next.
Expanding means meeting new people. AI decisions no longer sit with IT alone. At IAMCP earlier this year, Microsoft told Partners that business leaders now control around three quarters of AI services spend, which we covered in the four challenges your Copilot content should tackle. You’ll need conversations with operations and finance leaders as well as the IT team.
It also means partnering up. AI transformation is as much about people as technology, and change management is a discipline in its own right. Per Werngren, five-time worldwide president of the IAMCP, calls it competitive collaboration: specialise, team up with someone who does what you don’t, and tell the customer. “CIOs love this message,” he told us in our Original Thinkers interview. “They hate finding out there’s a subcontractor they never knew about.”
His rule of thumb: the lead partner on one deal becomes the supporting partner on the next.
Content is your differentiator
Here’s the catch. This is the third Microsoft room we’ve sat in over the past few months where Partners got the same brief. At IAMCP in London, Ferdi Van Huffel, Microsoft’s market intelligence lead across EMEA, named the four problems that stall Copilot rollouts. Kwadwo Benko, who led Microsoft’s internal Copilot adoption, told Partners to become customer zero. Now David, who shapes joint go-to-market strategies with Microsoft’s strategic partners, has mapped where customers get stuck.
Every Partner got the same memo. Repeating it won’t set you apart. Saying it through your own experts and your own customers’ stories will.
And the field is wide open. SmarterMSP found MSP leaders talk about AI on their personal LinkedIn profiles, while their company pages rarely carry consistent AI positioning or campaigns tied to the brand. When it comes to AI positioning for MSPs, most of your competitors haven’t joined up their message yet. smartermsp
Turning Microsoft’s framework into a message that sounds like you, and nobody else, is our job. As journalists by training, we interview the people in your business who’ve done the work, and find the detail a buyer will believe. Here’s what that sounds like at each stage:
- Shadow AI: “Your people are already using AI. Let’s make it safe.” A plain-English blog plus a tenant checklist, written for the MD, who’ll be the one alarmed by it (usefully).
- Intent: “Who should use AI first, and what should they get out of it?” Role-based advice for the operations and finance leaders who now hold the budget. We’ve covered how in selling Copilot to the C-suite.
- Proof of concept: “Which one process would you fix first?” An explainer from the consultant who gets pilots past the board, and a Q&A built from the questions your account managers hear every week.
- Periphery and core processes: “What would this look like redesigned around AI?” Case studies with a real before and after, measured in money or capacity, with time saved dropped as the headline metric. No customer willing to go on record? Be your own case study, and try our Case Study Trial.
- Agentic: “Before you add agents, do you know who can access what?” Thought leadership on governance and multi-agent control, for the customers who are ready.
Each message opens the next. That’s land and expand, done through content. Your partnerships make content too: a joint webinar with your change management partner reaches both your audiences. As Per puts it, “Marketing should establish the initial relationship with your potential customer, and sales comes in at the end.”
How to pay for content
Microsoft wants more Copilot licences sold. More than that, it wants AI so embedded in your customers’ day that, like David’s user, they can’t imagine working without it. And it’s paying for that. “If you want to be profitable with Microsoft, make sure you’re signed up for the incentives,” David said. He added that Microsoft has put more into its incentive programmes this year than ever before.
Co-op is part of that pot, and FY27 rewards demand generation. Under the Paid Media with Multi-Touch Campaign rules, content qualifies when it runs as one campaign: paid media at the centre, plus at least two other activities. Our FY27 Co-op guide has the detail.
Here’s a shadow AI campaign that would qualify:
- A blog, written from an interview with the engineer who keeps finding shadow AI in customer tenants.
- The tenant checklist, on a landing page, so you can see who’s interested.
- A customer zero case study on how you got AI under control in your own business.
- LinkedIn posts drawn from all three and boosted for six to eight weeks. That’s your paid element.
For the claim, you’ll need a screenshot of the paid activity, links to two pieces of creative and LinkedIn’s platform report. We run the paid side from your own ad account, send you the report monthly and itemise every invoice. It’s the kind of three-month push our Campaign Accelerator is built for.
Swap shadow AI for any other stage and the shape holds.
Meet them where they are
Microsoft has told every Partner what to say. The winners will say it in their own voice, at the stage each customer has reached.
We’ll keep sitting in these rooms so you don’t have to. Already delivering Copilot? We’ll find the transformation story in your customer work. Not there yet? We’ll start with your own business as customer zero. See how we approach Copilot marketing, then let’s talk.
FAQs
How can an MSP stand out on AI when every Partner hears the same message from Microsoft?
Through specifics. Anyone can repeat Microsoft’s framework. Only you can show your own experts fixing a customer’s problem, with a real before and after. Pick the stage where your customers are stuck and the driver you’re strongest in, then build your content around proof only you have. Run it as a campaign under your company brand as well as your leaders’ LinkedIn profiles.
How should an MSP talk to SMB customers about shadow AI?
Start with curiosity before alarm. Most employees using personal AI tools are trying to save time, and the business usually has no idea. Offer to find out what’s in use, explain the data risk plainly, and show a safer route through tools they already license. It’s a low-risk first conversation that often leads straight to the intent stage.
Can we use Microsoft Co-op funds to market our Copilot services?
Yes, if the content runs as part of a multi-touch campaign. Under FY27 rules, that means paid media plus at least two other activities, such as blogs, a landing page, webinars or a case study, planned together for one audience. A standalone blog or a one-off ad won’t qualify. Check your claim window and allocation in Partner Center first.
Should we be writing about agents yet?
Only for customers who are ready. For most SMBs, content about multi-agent control will go over their heads. A short piece on what agents will need from them, such as well-governed data and clear access controls, helps them prepare without getting ahead of themselves.
