llms.txt Intelligence

Publishing an llms.txt tells AI engines what your brand wants them to know. llms.txt Intelligence shows whether that declared content is actually shaping the answers AI engines give about you.

Ultra feature. llms.txt Intelligence is available on Ultra and Enterprise plans.

Before you begin

You need two things in place:

• An llms.txt published on your domain. Generate one if you don't have one yet.
• Prompt monitoring set up, so there are real AI answers to measure against. See Set up prompt monitoring.

Vizzybl ingests your declared content automatically once both are in place.

What it measures

Two scores, kept separate on purpose:

• Influence score — the share of your declared entries that appear in AI answers. This is a topical signal: your content is surfacing, but surfacing alone doesn't prove an engine read your llms.txt.
• Attribution score — the causally-defensible signal: the share of declared entries whose exact URL an engine actually cited. A cited declared URL rules out coincidence.

Read the attribution score as the firmer number and the influence score as the broader one.

The dashboard

Open llms.txt Intelligence from the sidebar. For each brand you'll see:

• Influence trend — how your influence score moves over time.
• Declared entries — every entry in your llms.txt and whether it lands in AI answers.
• Engine coverage — which AI engines surface each declared entry, as a grid of entries against engines.
• Dead weight — declared entries that never appear. Strengthen their descriptions or remove them.
• Answer gaps — brand pages AI engines cite that you haven't declared. Add these to your llms.txt.
• Causal control — whether your declared content lands more than comparable pages you didn't declare. A lift near zero means your llms.txt isn't adding influence yet.
• Competitor comparison — how your declared content performs against competitors', measured on the same monitored answers.

Reading small samples

AI answers vary between runs. When too few answers have been analyzed, the dashboard shows presence or absence instead of rates, and every rate carries a confidence interval. Don't read a trend into a handful of answers.

Refine your llms.txt

The Refine panel turns the gaps above into concrete edits — pages to add, entries to strengthen, topics competitors win on — and drafts a revised llms.txt. Copy or download it, publish it at your llms.txt URL, and Vizzybl re-measures on the next cycle.

That's the loop: measure influence, close the gaps, publish, measure again.

Next steps

• LLMs.txt Generation — Create or update your llms.txt files
• Citation Patterns (Ultra) — Understand why AI engines cite the sources they do
• Identify Your Competitors — Add competitors so the comparison has something to measure against