Win the user first: the channel should ride the pull, not fight it

Christian Lund, co-founder of Templafy, a document-generation software company, said in an interview with Raconteur published on 17 September that the way large companies buy software has flipped. For years the vendor’s job was to persuade the senior buyer who chose on behalf of the whole business. IT rolled the product out and staff were expected to use it. Lund told Raconteur that this changed almost overnight, and that employees now drive the decision.

His rule for vendors runs to seven words: “You have to get the user first.” Lund told Raconteur that the employee wants to work fast and the employer wants control, and that a vendor has to satisfy both. Templafy, he said, decided early that the business would need a way into the conversation between the user and the tool, which is why its product carries the company’s templates and compliance rules inside the software the employee actually uses.

He also refused to treat speed as the prize. Lund told Raconteur that saving time has become close to a commodity, that a faster worker is not automatically a more productive one, and that plenty of AI tools impress in a demonstration and fail on a real use case. And Templafy does not build custom products for a single customer: a change has to help many clients or it is not made. The company says more than 4 million knowledge workers use its platform, which is the scale at which that discipline is affordable.

I agree with the diagnosis. I would go further: “win the user first” is also the mechanism by which shadow AI and unmanaged SaaS spread through UK companies, and the channel has spent the past few years treating that pull as a security problem when it is one of the largest pieces of unserved demand in the market. Lund is building a product around the same force that walks past the reseller, the distributor and the license.

The user has already won

Deloitte’s survey of 25,000 UK workers, published on 16 September with fieldwork by Ipsos UK in May and June 2026, puts numbers on it. Deloitte found that 63% of working adults in the UK use generative AI (GenAI) for their jobs. Nearly a third (31%) of those users do so without their employer’s knowledge. Between them, British workers spend £958m a year of their own money on GenAI tools for work. Almost half (46%) rely on free tools, 17% pay for their own, and roughly half have had no formal training.

Hayley McKelvey, chief AI officer at Deloitte UK, said in the release: “UK workers are showing they don’t want to wait for permission to use GenAI.” That is Lund’s thesis, measured. The user got there first, paid for it and did not ask. Nearly £1bn a year is the size of a market the channel is not in.

Blocking the pull is a losing trade

The channel’s reflex has been to sell the CISO something that says no: a cloud access security broker (CASB), data loss prevention (DLP), a browser policy. Those are useful as one layer. As a strategy they fail, because they treat demand as a threat. An employee told no on the corporate laptop opens a personal phone. Deloitte’s 31% is what “no” looks like at scale: the tool is still in use, the data is still leaving, and nobody has a license, a log or an invoice.

The better reading of the same number is that the discovery phase of the sale is already complete. Every partner has been told to find the business problem before pitching the product. Here the users have found it, chosen a tool and paid for it themselves. What is missing is a governed, licensed version of what they are already doing. That is a channel product.

Turn pull into a licensed deployment

Four moves follow, and none of them requires a new vendor.

Sell the discovery. A fixed-fee audit of which AI tools are in use, in which teams and with what data is a conversation the CIO wants to have and the CFO will fund. Its output is a list of subscriptions to consolidate, which is the pipeline for everything below.

Convert personal spend into corporate agreements. The £958m in Deloitte’s survey is paid on personal cards for consumer tiers with consumer terms. Moving those users onto enterprise licenses with data controls turns spend the company cannot see into a contract it can negotiate, and it ends the leakage. That is a margin line and a governance win on the same order form.

Bundle the training. Roughly half of GenAI users have had none, according to Deloitte. Training is a service the channel can price, and it is what makes a CIO comfortable signing the agreement above.

Pick vendors who did what Lund describes. The test is simple: can an administrator set a policy without the user noticing anything but the result? If the controls live in a console the user never sees and the experience gets worse, staff will go back to the free tool and the deployment will fail from below.

Where the channel should bend

The one place I part company with Lund is his refusal to build for a single customer. That rule is right for a vendor with millions of users. It is wrong for a partner, because bending is the partner’s product. A vendor sells the standard; the channel sells the configuration, the integration and the process wrapped around it, one company at a time. Partners should not copy Templafy’s discipline. They should be the custom layer that vendors like Templafy have chosen not to be, and charge for it.

Lund is right that the CIO no longer signs the first order. The user does, often on a personal card. The channel’s job is to make sure the company signs the second one.