The AI gap your clients are asking about that you cannot deliver yet

A client asks the question over a renewal call or at the end of a support ticket, almost in passing. Can you help us with AI? Can you get us going with Copilot? Can you automate this thing we keep doing by hand? It is a reasonable question to put to the firm that already looks after their technology. The awkward part is the honest internal answer, which for a lot of partners right now is some version of not really, not yet. The demand has arrived faster than the ability to meet it, and the gap between the two is worth looking at squarely, because the way a partner responds to that question over the next year or two is a commercial decision, not just a technical one.

The question clients are already asking

The demand is no longer speculative. How many businesses are actually using AI depends on how it is counted, but every recent UK measure points the same way and the trend is steep. The British Chambers of Commerce found 35% of SMEs actively using AI in 2025, up from 25% a year earlier, while the share with no plans to adopt fell from 43% to 33% 1. A separate YouGov poll of UK SME leaders put current use at 31%, with another 15% planning to adopt 3. The government’s own research, which uses the most conservative definition across all business sizes, puts current use at 16%, but rising with company size: 14% of micro businesses, 23% of mid-sized, and 36% of large 2. So depending on how it is counted, somewhere between roughly one in six businesses and one in three SMEs is already using AI in some form, and the rest are watching.

For a partner, the more telling figure is how many clients are at least asking. In OpenText’s 2025 survey of managed service providers, 95% reported customers either exploring or actively adopting Microsoft Copilot 4. Almost nobody is being left alone on this. The question is arriving from nearly every client, whether the partner is ready for it or not.

Why the gap is real, not nerves

The instinct is to read partner hesitation as caution that will pass. The numbers suggest it is something firmer than that. In the same OpenText research, confidence in being ready to support AI-related needs fell from 90% in 2024 to under half in 2025, as the question moved from interest to actual deployment, and the report notes that fewer than half of providers feel fully confident guiding customers on AI tools 4. That is a managed-security survey, so the confidence figure sits in a security context, but the direction is striking: the closer the work got to real, the more the confidence fell.

The wider channel data tells the same story. The Channel Company found 83% of solution providers taking a wait-and-see posture on AI investment in 2025, up from 78% the year before, with the aggressive investors shrinking from 22% to 17% 5. In separate research commissioned by Westcon-Comstor, 74% of partners said they were not yet able to design and deliver AI-ready solutions despite strong customer demand 6. That last study is vendor-commissioned and framed around AI networking, so it is best read as illustrative of the channel pattern rather than a precise figure, and like the OpenText and Channel Company numbers it is global channel research rather than a UK-specific measure. Taken together, though, the pattern is hard to miss. The demand side has moved and the supply side has not caught up.

The bottleneck is not appetite. It is skills and clear use cases. The government’s research found 60% of businesses citing limited AI skills, expertise and knowledge as a barrier, and 71% citing no identified need or use for AI yet 2. A client asking can you help us is often a client who does not know what they want help with, which makes the question harder to answer well, not easier. Saying yes credibly means being able to find the use case, scope it, and deliver it, and that is precisely the capability most partners have not built yet.

The trap is that reselling is the worst-paid option

Faced with that gap, the obvious move is to resell an AI platform: sign up as a reseller, add a line to the price list, pass the licences through. It feels like progress, and it is the lowest-effort way to answer the client’s question with a yes. It is also, on the channel’s own analysis, the worst-paid way to do it.

The reason is simple. The licence revenue flows to the hyperscalers and the large software vendors who own the platforms. The partner is left doing the integration and the day-to-day management for thin margins, while the recurring value accrues to someone else 7. The money in AI for a partner is not in the resale. It is in the work around the resale: working out what a client should actually do, integrating it into how they run, and governing it once it is live. McKinsey’s analysis of AI-enabled customer service, for instance, points to a 40% to 50% reduction in service interactions and more than a 20% reduction in cost-to-serve in the transformations it studied 8. Those are enterprise figures, not a promise of SME outcomes, so they are best read as directional evidence of where the value sits rather than a number to quote at a client. The value is real, but it lives in the outcome, not the licence. Reselling captures the least of it.

The three honest options

If reselling is the weak answer, what are the real ones? There are broadly three, and which fits depends on a partner’s size, ambition, and the relationships it wants to keep. None of them is the single right answer.

The first is to build fluency before selling. Close the skills gap internally, get genuinely confident with the tools, find the use cases in a few trusted accounts, and learn what good delivery looks like before promising it at scale. This is the slowest route and it costs real time, but it is the most defensible, because it builds capability the partner owns. The risk is that the demand does not wait for the learning curve.

The second is to move up the stack. Stop thinking of AI as a product to resell and start thinking of it as advisory, integration, and governance work to deliver. This is where the channel research consistently locates the margin, and it plays to what a partner already has: the client relationship, the knowledge of how that business actually runs, and the trust to recommend. It asks the partner to sell judgement rather than licences, which is a harder thing to package but a more durable one to own.

The third is to partner for the delivery a partner cannot yet staff. Bring in a specialist for the build, keep owning the client relationship, and present the outcome rather than the supply chain behind it. This answers the client’s question with a real yes today, without overpromising on capability the partner does not have in-house yet, and without ceding the relationship. It works only where the specialist is content to sit behind the partner rather than in front of the client, which is the whole basis of a genuine white-label arrangement.

These are not mutually exclusive. A sensible partner often does all three at once: partnering for delivery now, moving up the stack into advisory over time, and building internal fluency on the back of the work. What matters is choosing deliberately rather than defaulting to the resale because it is the easiest box to tick.

A timing problem, not a permanent one

The gap between what clients are asking for and what most partners can deliver is real, but it is a timing problem rather than a permanent state. The demand has crossed over and is not going back; the capability is catching up and will. The partners who come out of this period well are not the ones who had all the answers first. They are the ones who answered the client honestly in the meantime: clear about what they could deliver themselves, candid about what they could not yet, and willing to bring in help rather than overpromise. Answering can you help us with AI? with a confident, honest yes, backed by real delivery, is worth far more to a long relationship than a thin reseller margin.

For partners weighing the third of those options, where a specialist handles the delivery and the partner keeps the client, the site goes into how that arrangement works in more detail.

See how hirevolution works with partners