Agentic AEO · How it works

How Agentic AEO works

I build a team of AI agents on Grok Bot that measures how often AI assistants cite you, finds what to fix, and designs, writes and critiques the changes, then builds them. Research, design, copy, critique and build are separate roles, and measurement is independent of research, so no agent marks its own homework. I run it, you approve the work once a week, and at the end I hand the system over.

The flow in five steps

  1. 1

    Onboarding

    Agree your customers, competitors and buyer questions, then record a fixed prompt set and a baseline.

  2. 2

    Measure and research

    Measurement runs those prompts across ChatGPT, Perplexity and Google AI Overviews, several times each. Research finds what to fix and which pages engines already cite.

  3. 3

    Design, copy and sign-off

    The designer and copywriter produce the work. A critic reviews it once, then you sign it off in the weekly batch. Schema, FAQs and meta descriptions can take a pre-agreed fast lane.

  4. 4

    Build or export

    Preview pages are built, and I approve them before anything ships. No CMS access? Your developers get the code, instructions and a changelog.

  5. 5

    Re-check and hand over

    Two to four weeks later the same prompts are checked again, then I hand the system over. Rolling monthly support if needed.

The agent team on Grok Bot

White-glove where it matters: I sit outside the team as Account Director, and off-site work is run by a person, not an agent.

  • Account Director: me, backed by my Chief of Staff agent

    I sit outside the agent team. I manage the agent workflow, escalate issues, run the weekly batched sign-off, and send outcome reports on enquiries and citations. I also own plan and usage management.

  • Onboarding and discovery

    The first step, before any research or site changes. Builds the ICP and competitor set, and pulls real buyer questions from sales calls, support tickets and customer interviews. The output is a fixed prompt set and a baseline.

  • Measurement agent

    Independent of Research, so the same agent doesn't mark its own homework. Runs the fixed prompt set across ChatGPT, Perplexity and Google AI Overviews, several times each, because answers vary. Tracks AI referral traffic in GA4 and a “How did you hear about us” form field that includes AI assistants.

  • Research agent

    Identifies opportunities, finds the third-party pages AI engines already cite, and suggests fixes. Works with a licensed tracker (Peec) and official APIs. The browser is used only for one-off checks.

  • Off-site work (managed by me)

    A parallel track, managed by a person rather than an agent. Uses platforms like That's Noble across Reddit, review sites, listicles, PR, YouTube and LinkedIn, to earn mentions on the pages AI engines actually cite.

  • Web Designer agent

    Turns agreed fixes into concept pages, then hands them to the Copywriter, and revises once after critique.

  • Copywriter agent

    Writes and tightens the copy after the design, and revises once after critique.

  • Critic agent

    Critiques the design and copy, as a sceptical HR tech buyer and an AI answer engine, for one revision loop, then sends the final with a before and after summary to me for your weekly sign-off.

  • Web Developer agent, or a code export

    Creates preview pages, I approve them, and the agent pushes live within the guardrails for brand voice, legal and compliance sign-off, and rollback. Or, without CMS access, the work is exported as code for your developers, with instructions and a changelog.

  • Low-risk fast lane

    Pre-agreed rules let schema, FAQs and meta descriptions ship without a separate approval, so routine technical changes aren't stuck in the weekly batch.

Weekly sign-off and pace

After one critique loop, you review and approve work once a week, not piecemeal. Low-risk fixes take a pre-agreed fast lane.

You get your own Grok Bot instance. The pace of work is set by that instance's compute, not by an open-ended retainer of agent hours. Standard pace is the $60 a month plan, and work is scoped to stay within it. If you want to go faster, you can move to the next plan up at $200 a month. I agree any upgrade with you before the pace changes, and I report usage monthly.

Independent measurement

Against the month 1 baseline, with one measure for each step in the chain.

  • Citations and visibility. How often you're mentioned or cited for the fixed prompt set, across each engine.
  • AI referral traffic. Visits from AI assistants, tracked in GA4.
  • Enquiries attributed to AI search. Through GA4 and the “How did you hear about us” form field.
  • Opportunities created. Reported by your team from those enquiries.

See it on my own site

Free discovery call

Start with a free discovery call

A free 30-minute call. Together, on the call, we look at whether buyers in your category use AI to shortlist, whether you show up, and whether you can act on the fixes. If it isn't worth doing, I'll tell you. The £595 audit and the 3-month service both start here.

  • Where you're named and cited, and who's named instead
  • The pages AI engines already rely on
  • What to fix first