AgentPratham

BENCHMARK 2026

We scanned 1,659 UK websites to see which are ready for AI agents.
Most are not.

The full benchmark report scores UK websites across the three pillars that decide whether AI agents can read, trust and act on them. Read the findings, see where your clients sit, and see what good looks like.

The Benchmark

The UK AI Readiness Report 2026.

1,659 businesses. 11 verticals. One score. The first benchmark of AI visibility across the business landscape.

30%
not visible to AI agents
1,659
websites scored
11
verticals covered
15%
have adopted llms.txt

What we measure

The PARSE Score: three core pillars.

100 points across three core pillars, measuring whether your site can be reached, interpreted, and trusted by automated clients.

01 / Pillar

Technical Foundation

The infrastructure, performance, and configuration that determines how reliably AI agents can access and process your site.

02 / Pillar

Semantic Truth

The accuracy and clarity of your content, and how well automated systems can extract, interpret, and trust what your pages say.

03 / Pillar

Actionable Handshake

Structured signals and directives that tell AI systems what your site offers and what interactions are available to automated clients.

Coming Soon

There is also a tool.

We are building the audit behind the benchmark into something agencies can run for any client site. It is not open yet. A small group of agencies is getting early access before public launch.

See what it does

QUARTERLY BENCHMARK

New benchmark data every quarter.

Get it before it goes public.

No spam. Research updates only.