AI assistants at the moment are a discovery channel. Individuals who used to sort a question into Google now have the choice to ask ChatGPT, Claude, Gemini, Perplexity, or Copilot as a substitute. And the reply they get again names just a few manufacturers, cites just a few sources, and strikes on.
Â
If you happen to’re one of many named manufacturers, you win. If you happen to’re not, you by no means knew the dialog occurred.
Optimizing for that’s referred to as AEO (or for those who desire, GEO.) That is now a part of search engine optimization and/or the advertising and marketing workforce’s duty.
The unlucky drawback is that it’s relentless. The solutions shift weekly if not each day. The solutions fluctuate by mannequin, by nation, by phrasing. In actual fact, the identical immediate generates different brands in different lists all the time.
AI assistants are extremely inconsistent in terms of recommending manufacturers or merchandise. Suggestions change each time you ask the identical query.
A competitor you’ve by no means heard of might begin getting cited and also you solely discover out a month later. Or someplace in a Reddit thread that an LLM retains quoting, your pricing is improper.
None of this journeys an alarm. However you want somebody checking, on a schedule, commonly. It doesn’t must be you. It simply must ping you when one thing’s value your consideration.
That is the a part of AEO you may hand to Letaido.
Listed here are a number of the greatest use circumstances you will get began with.
What’s Letaido?
Letaido is a advertising and marketing agent from Ahrefs—an AI assistant with direct entry to the total Ahrefs dataset that may perform advertising and marketing duties autonomously, quite than simply reply questions.
Letaido consists of:
- Unrestricted entry to Ahrefs endpoints. Each endpoint we use to construct Ahrefs is offered, together with many you can not attain by way of API or MCP.
- Critical tech stack beneath. Postgres for state, Flask for UIs, an OpenRouter proxy with 300+ fashions, net fetch with full-page parsing, PDFs, OCR, scheduled jobs.
- Native connectors to advertising and marketing instruments. Slack, HubSpot, GitHub, Notion, Linear, Mailchimp, Resend, SendGrid, Stripe, Gong, WordPress, Airtable, Apify, and even Semrush.
- Skilled talent library. The Ahrefs workforce has contributed pre-built advertising and marketing expertise and purposes that encode how we truly work.
If you do key phrase analysis, your objective is to search out what folks typed into Google’s search field.
AEO is the same-same-but-different job: you’re attempting to determine which prompts and questions folks now put to an assistant, and which of them are value profitable.
Sadly, you may’t measure immediate quantity instantly. However you may estimate it.
Right here’s how:
Give Letaido just a few seed subjects and it makes use of Model Radar to floor the actual prompts and questions folks ask in your area of interest, then pulls the search demand behind each as a proxy for a way typically it’s requested.
Since uncooked search quantity overstates immediate demand (as a result of fewer people use each AI assistant than Google), Letaido scales every immediate’s demand by the assistant’s person base relative to Google Search.
For instance, if ChatGPT has roughly 30% of Google’s customers, a immediate’s Google quantity is weighted by ~0.3 to replicate its actual AI demand.
What comes again is a prioritized map of prompts, ranked by the search demand behind each (a tough proxy) and the way industrial the intent is.
You find yourself with a shortlist of the prompts that truly matter in your area, as a substitute of guessing.
Starter immediate
Construct me an AEO prompt-discovery instrument. Enter: just a few seed subjects. (1) Use Ahrefs Model Radar to floor the actual questions folks ask on this area of interest and pull their search demand; (2) cross-check quantity and intent towards Key phrases Explorer; (3) rating every immediate by search demand (as a proxy for immediate demand) Ă— industrial intent; (4) cluster the prompts into themes. Output a ranked immediate map in a desk I can export, with quantity, intent, and theme per immediate.
Everybody needs to know “are we displaying up in ChatGPT?”
However the factor is: it’s completely different for each platform. Even for those who’re displaying up on ChatGPT, it doesn’t imply you’re doing the identical for AI Overviews, AI Mode, and extra.
Right here’s how one can see your actual share of voice.
Ask Letaido to measure your actual share of voice throughout the assistants that matter to you (e.g., ChatGPT, Google AI Overviews, AI Mode, Gemini, and so on.). It runs your precedence prompts by Model Radar, counts how typically every reply mentions or recommends you versus your named opponents, and breaks it down immediate by immediate so you may see precisely which of them you win and which you lose.
The output is a scoreboard: your share of voice per platform, the prompts the place you’re absent, and the opponents consuming your share.
Starter immediate:
Starter immediate
Construct me an AI share-of-voice tracker. For my listing of precedence prompts and named opponents, use Ahrefs Model Radar to measure my share of voice throughout ChatGPT, Perplexity, Gemini, Copilot, Grok, and Google AI Overviews/AI Mode. Break it down per platform and per immediate: the place I’m talked about, the place I’m advisable, the place I’m absent, and which competitor wins as a substitute. Log the weekly totals so I can observe the pattern, and solely ever present my share relative to my named opponents.
When an assistant recommends a competitor as a substitute of you, it’s as a result of it trusted particular sources. These sources aren’t a thriller. They’re listed, proper there within the citations.
Ask Letaido to reverse engineer them.
For the prompts you care about, it makes use of Model Radar to drag the precise domains and pages AI cites when it solutions — the assessment websites, the listicles, the docs, the discussion board threads — and ranks them by how typically they’re cited and the way a lot authority they carry.
Now you might have a goal listing: earn your manner into these particular sources (a point out, a list, a correction, a visitor piece) and you modify what AI says subsequent time.
That is the place quite a lot of AEO work truly occurs, as a result of most model mentions that form AI solutions come from different web sites, not yours.
Starter immediate
Construct me a citation-source finder. For my precedence prompts, use Ahrefs Model Radar to drag the precise domains and pages AI platforms cite when recommending my opponents. Rank them by quotation frequency and authority, mark which of them already point out me vs. don’t, and label every with the play to earn in (get listed, get reviewed, request a correction, pitch a contribution). Output a prioritized outreach listing.
Assistants don’t simply determine whether or not to say you. In addition they determine how one can body you. And that framing shifts continually throughout platforms and prompts. You desire a standing learn on whether or not AI talks about your model positively, neutrally, or negatively, and the place the unfavorable framing is concentrated.
Right here’s how one can get Letaido to deal with it.
Get Letaido to drag the total textual content of each AI reply about you from Model Radar, rating the sentiment, and observe the totals over time. The output is a scoreboard: your share of constructive versus unfavorable framing per platform, the prompts the place the framing is worst, and the pattern because it strikes week to week.
That is the standing audit that tells you whether or not you might have a notion drawback and the place.
Then comply with Despina’s information on auditing model mentions to be taught what to do subsequent.
Starter immediate
Construct me a negative-sentiment tracer. Use Ahrefs Model Radar to drag the total reply textual content, immediate, and cited sources for each AI point out of my model throughout platforms. Run sentiment evaluation on every reply, preserve the unfavorable ones, and for every: quote the road, identify the triggering immediate, and establish the cited supply most definitely accountable. Log them to Airtable sorted by severity, and re-run weekly so I can watch points clear as I repair the sources.
Assistants hallucinate confidently. They’ll declare you might have a function you killed two years in the past, quote a value that was by no means actual, or attribute a competitor’s integration to you…. in fluent, authoritative prose
Left alone, that turns into what hundreds of consumers imagine.
Give Letaido your floor fact: your docs, your pricing web page, your function listing, and so on. Then get it to fact-check the AI assistants by pulling each AI reply about you from Model Radar, verify every factual declare towards your supply of fact, and flag the fabrications.
For each, it traces the cited supply that possible seeded it, so whether or not to repair a web page, request a correction, or feed a greater sign.
The result’s an inventory of issues AI is improper about you, every with the receipt.
Starter immediate
Construct me an AI hallucination catcher. Inputs: my source-of-truth docs (pricing, function listing, product pages) and my model. Use Ahrefs Model Radar to drag each AI reply about me throughout platforms, extract every factual declare, and verify it towards my docs. Flag each declare that’s false or outdated, quote it, present the contradicting floor fact, and hint the cited supply that most definitely triggered it. Output a repair listing grouped by supply so I can right the worst offenders first.
Each AEO program finally has to justify itself to somebody who doesn’t care about citations, solely outcomes. That report is tedious to assemble by hand and simple to place off, so it by no means will get made.
Get Letaido to jot down it.
On a schedule, it stitches collectively your AI share of voice and quotation developments from Model Radar, your search efficiency from Search Console, and your AI assistant referral site visitors from Internet Analytics into one client-ready month-to-month narrative.
KPI tiles, month-over-month motion, the prompts you gained and misplaced, and a brief written abstract a non-specialist can comply with.
Starter immediate
Construct me a month-to-month AI-visibility report. On the first, pull my AI share of voice and quotation developments from Ahrefs Model Radar (ChatGPT, Perplexity, Gemini, Copilot, Google AI Overviews/AI Mode), my Search Console efficiency, and my AI-assistant referral site visitors from Ahrefs Internet Analytics. Compose KPI tiles, month-over-month deltas, prompts gained/misplaced, and a plain-English abstract into one shareable report, and publish a hyperlink to my #advertising and marketing Slack channel.
Closing ideas
AEO isn’t a mission you end. The solutions transfer each week, and the protection is having one thing watching while you’re not.
That’s the actual unlock right here: Get Letaido to run all of those on a schedule and ping you when a competitor begins profitable a immediate, when your pricing exhibits up improper, and when sentiment slips.
If you happen to’re an Ahrefs buyer, you may strive Letaido at no cost for one month.
Paste a starter immediate right into a contemporary workspace and also you’ll have the primary model working right now. Then refine and customise it to your required workflow. You’ll have your private AEO instruments in no time.
