On 27 October the Inneall team spent the day at Abbey Road Studios in London for Progress MartechNEXT 2026, the London finale of the MartechNEXT world tour.

It was a fitting venue. Abbey Road opened in 1931 as the world’s first purpose-built recording studio, and as Sara Faatz of Progress reminded everyone in her opening remarks, it has always been a place for experimenting with technology in music. EMI engineer Alan Blumlein, who patented stereo sound a month after the studios opened, made some of the first stereo recordings there. Its engineers went on to build early mixing desks and echo chambers and invented artificial double tracking, and today the studios run Europe’s first music-focused technology incubator. A room with that history was a good place to ask what AI will do to marketing.

AI dominated the agenda, as expected. What made the day useful was how practical it was. These are the ideas we think matter most for the colleges and organisations we work with, and what we are doing about them.

Three kinds of AI agent

Scott Brinker, editor of chiefmartec, framed AI agents in marketing as three distinct types: agents that work for marketers, agents that serve customers, and agents that act on behalf of customers.

The first two are familiar: assistants that draft content and analyse campaigns, and chat assistants on a website. The third is the one we kept coming back to. When a prospective student asks an AI assistant to find part-time business courses in Dublin that start in January and cost under a certain amount, the agent does the research, compares the options and comes back with a shortlist. The college may never see the visit that led to the decision.

The website matters more, not less

It would be easy to conclude that websites matter less in an agent-led world. The view in the room was the opposite. The website remains the foundation of an organisation’s digital presence and the one source of trusted information about its programmes, services and expertise. Agents need something reliable to read.

For a college, that changes the brief for course pages:

• Accurate: fees, start dates, entry requirements and delivery modes that match what admissions will actually tell an applicant.
• Structured: course data that machines can read, not only paragraphs written for people.
• Current: no last year’s fees or a closed intake still showing as open.
• Connected: course information drawn from the systems that hold the real data, rather than retyped by hand.
• Still built for people: when an applicant does visit, the experience has to earn their trust and make it easy to apply.

Search without the click

Theresa Williams and Natalka Antoniuk of Appius took on the realities of SEO and GEO (generative engine optimisation) in a world where fewer people click through. Their message was not to panic about lost traffic. Much of that traffic was never going to turn into a sale or an application, and discovery now happens largely away from your website, inside search platforms, answer engines and large language models.

What wins in that world is what has always built reputation: valuable, honest, credible and consistent content, and real brand authority. The other half of the argument was measurement. If sessions fall while enquiries hold steady, that is not failure, but you need to measure what really matters to show it, so leadership keeps backing the approach.

Where Sitefinity is heading

For us as a Sitefinity partner, the most relevant session was Progress setting out how digital experiences will evolve with AI. It follows the direction we saw at the Partner Roundtable in Barcelona earlier this year, and it is now concrete enough to plan around.

The practical point for existing Sitefinity sites is that each step depends on the one before. Conversational answers are only as good as the content underneath them, and personalisation needs clean, structured data. The work that makes a site ready for AI agents is the same work that makes these features worth switching on.

Build what you need

Brinker’s other theme was the rise of custom-built apps and agents. As AI makes software easier to build, marketers can create small tools that solve their own problems, faster and with less dependence on IT, instead of adding another product to an already crowded stack. His point, that the people closest to a problem often understand best how to solve it, is one we recognise from our own work.

We agree, with one condition: governance. We run our own company on the AI we sell, and every agent we build has its own login, named tools, limited permissions and a person approving changes. That is what turns a clever prototype into something a college can trust with real data.

The day in pictures

What we are bringing back to our clients

Events are only useful if something changes afterwards. These are the five things we will be raising with the colleges and organisations we work with over the coming months.

What to do next

If you run a Sitefinity site, or any college website, and want to know how it looks to AI agents and answer engines, talk to us. We will review what AI assistants currently say about you, how your course data is structured and what it would take to get your site ready for what Progress is building next.