Overhead view of three colleagues reviewing design work at a table

•

3 minute read

The Government Consultation Closing This Month That Most Agencies Haven’t Noticed

Every agency I work with runs some kind of scheduling or booking system. Freelancers, crew, editors, whoever's next in the rota. Increasingly, that system doesn't just record who worked when. It starts making small calls on its own: who gets offered the next job, whose reliability score nudges up or down, who shows as available first. Nobody sat down and designed that as "algorithmic management." It's just the booking software everyone has used for years, quietly updated with a bit of AI to save someone doing the allocation by hand.

That's precisely the territory a UK government consultation opened on 8 July 2026. It closes on 30 September, three weeks from today. Reporting from Ashurst and AO Shearman's employment team, and the consultation document itself on GOV.UK, makes clear this isn't limited to the familiar territory of attendance tracking and email monitoring. It explicitly covers AI-driven algorithmic management and fully automated decisions about workers, and it could lead to draft legislation or a statutory code of practice.

Nothing is law yet. That's worth saying plainly, because the point here isn't to create alarm about a deadline that doesn't actually require agencies to do anything by 30 September. The point is narrower and more useful: this is the first clear sign that "the software decides who gets the next booking" is moving from an operational detail nobody thinks twice about, to a category regulators are actively defining.

Agencies are more exposed to this than most SMEs, not because they're doing anything wrong, but because of how they're structured. A 40-person events or production agency might have five staff and thirty-five freelancers cycling through jobs, and the tool coordinating that is almost certainly making some kind of automated allocation decision under the surface. That's a very different starting point from a business where every scheduling decision still runs through a manager's judgement. It's not a reason to panic. It's a reason to actually know what the tool is doing.

That's the useful exercise here, and it doesn't require waiting for the consultation to close or for legislation to appear. It means sitting down with whatever scheduling, booking, or performance software is already in daily use and asking a short, specific set of questions. What decisions does it make without a person checking first? What data is it using to make them? If someone challenged one of those decisions tomorrow, could anyone explain why it happened? Most agencies have never asked, because the software arrived gradually and nobody flagged the moment it started deciding rather than just recording.

This is exactly the ground the Governance dimension of the AI Accelerator Diagnostic is built to cover. Not a compliance audit for its own sake, but a plain answer to the question of what everyone in the business needs to know before using AI in real work, including the AI that's already quietly running in the background of tools nobody thinks of as "AI tools" at all. It sits alongside the wider shift already under way with the EU AI Act, in force since August 2024 and now moving through full enforcement, which touches any UK business with EU customers, employees, or operations regardless of Brexit. The domestic and European pictures are starting to rhyme.

Three weeks is not long enough to need a crisis response. It's exactly long enough to find out, calmly, what's already running in your business and whether anyone could explain it if asked. That's a better use of the window than waiting to see what the consultation produces and reacting after the fact.

Go wisely.