If you’re one of the named brands, you win. If you’re not, you never knew the conversation happened.
Optimizing for that is called AEO (or if you prefer, GEO.) This is now part of SEO and/or the marketing team’s responsibility.
The unfortunate problem is that it’s relentless. The answers shift weekly if not daily. The answers vary by model, by country, by phrasing. In fact, the same prompt generates different brands in different lists all the time.
AI assistants are highly inconsistent when it comes to recommending brands or products. Recommendations change every time you ask the same question.
A competitor you’ve never heard of could start getting cited and you only find out a month later. Or somewhere in a Reddit thread that an LLM keeps quoting, your pricing is wrong.
None of this trips an alarm. But you need someone checking, on a schedule, regularly. It doesn’t have to be you. It just needs to ping you when something’s worth your attention.
This is the part of AEO you can hand to Letaido.
Here are some of the best use cases you can get started with.
Letaido is a marketing agent from Ahrefs—an AI assistant with direct access to the full Ahrefs dataset that can carry out marketing tasks autonomously, rather than just answer questions.
Letaido includes:
- Unrestricted access to Ahrefs endpoints. Every endpoint we use to build Ahrefs is available, including many you cannot reach via API or MCP.
- Serious tech stack underneath. Postgres for state, Flask for UIs, an OpenRouter proxy with 300+ models, web fetch with full-page parsing, PDFs, OCR, scheduled jobs.
- Native connectors to marketing tools. Slack, HubSpot, GitHub, Notion, Linear, Mailchimp, Resend, SendGrid, Stripe, Gong, WordPress, Airtable, Apify, and even Semrush.
- Expert skill library. The Ahrefs team has contributed pre-built marketing skills and applications that encode how we actually work.
When you do keyword research, your goal is to find what people typed into Google’s search box.
AEO is the same-same-but-different job: you’re trying to figure out which prompts and questions people now put to an assistant, and which ones are worth winning.
Unfortunately, you can’t measure prompt volume directly. But you can estimate it.
Here’s how:
Give Letaido a few seed topics and it uses Brand Radar to surface the real prompts and questions people ask in your niche, then pulls the search demand behind each one as a proxy for how often it’s asked.
Since raw search volume overstates prompt demand (because fewer people use each AI assistant than Google), Letaido scales each prompt’s demand by the assistant’s user base relative to Google Search.
For example, if ChatGPT has roughly 30% of Google’s users, a prompt’s Google volume is weighted by ~0.3 to reflect its real AI demand.
What comes back is a prioritized map of prompts, ranked by the search demand behind each one (a rough proxy) and how commercial the intent is.
You end up with a shortlist of the prompts that actually matter in your space, instead of guessing.
Build me an AEO prompt-discovery tool. Input: a few seed topics. (1) Use Ahrefs Brand Radar to surface the real questions people ask in this niche and pull their search demand; (2) cross-check volume and intent against Keywords Explorer; (3) score each prompt by search demand (as a proxy for prompt demand) × commercial intent; (4) cluster the prompts into themes. Output a ranked prompt map in a table I can export, with volume, intent, and theme per prompt.
Everyone wants to know “are we showing up in ChatGPT?”
But the thing is: it’s different for every platform. Even if you’re showing up on ChatGPT, it doesn’t mean you’re doing the same for AI Overviews, AI Mode, and more.
Here’s how to see your real share of voice.
Ask Letaido to measure your real share of voice across the assistants that matter to you (e.g., ChatGPT, Google AI Overviews, AI Mode, Gemini, etc.). It runs your priority prompts through Brand Radar, counts how often each answer mentions or recommends you versus your named competitors, and breaks it down prompt by prompt so you can see exactly which ones you win and which you lose.
The output is a scoreboard: your share of voice per platform, the prompts where you’re absent, and the competitors eating your share.
Starter prompt:
Build me an AI share-of-voice tracker. For my list of priority prompts and named competitors, use Ahrefs Brand Radar to measure my share of voice across ChatGPT, Perplexity, Gemini, Copilot, Grok, and Google AI Overviews/AI Mode. Break it down per platform and per prompt: where I’m mentioned, where I’m recommended, where I’m absent, and which competitor wins instead. Log the weekly totals so I can track the trend, and only ever show my share relative to my named competitors.
When an assistant recommends a competitor instead of you, it’s because it trusted specific sources. Those sources aren’t a mystery. They’re listed, right there in the citations.
Ask Letaido to reverse engineer them.
For the prompts you care about, it uses Brand Radar to pull the exact domains and pages AI cites when it answers — the review sites, the listicles, the docs, the forum threads — and ranks them by how often they’re cited and how much authority they carry.
Now you have a target list: earn your way into those specific sources (a mention, a listing, a correction, a guest piece) and you change what AI says next time.
This is where a lot of AEO work actually happens, because most brand mentions that shape AI answers come from other websites, not yours.
Build me a citation-source finder. For my priority prompts, use Ahrefs Brand Radar to pull the exact domains and pages AI platforms cite when recommending my competitors. Rank them by citation frequency and authority, mark which ones already mention me vs. don’t, and label each with the play to earn in (get listed, get reviewed, request a correction, pitch a contribution). Output a prioritized outreach list.
Assistants don’t just decide whether to mention you. They also decide how to frame you. And that framing shifts constantly across platforms and prompts. You want a standing read on whether AI talks about your brand positively, neutrally, or negatively, and where the negative framing is concentrated.
Here’s how you can get Letaido to handle it.
Get Letaido to pull the full text of every AI answer about you from Brand Radar, score the sentiment, and track the totals over time. The output is a scoreboard: your share of positive versus negative framing per platform, the prompts where the framing is worst, and the trend as it moves week to week.
This is the standing audit that tells you whether you have a perception problem and where.
Then follow Despina’s guide on auditing brand mentions to learn what to do next.
Build me a negative-sentiment tracer. Use Ahrefs Brand Radar to pull the full answer text, prompt, and cited sources for every AI mention of my brand across platforms. Run sentiment analysis on each answer, keep the negative ones, and for each: quote the line, name the triggering prompt, and identify the cited source most likely responsible. Log them to Airtable sorted by severity, and re-run weekly so I can watch issues clear as I fix the sources.
Assistants hallucinate confidently. They’ll claim you have a feature you killed two years ago, quote a price that was never real, or attribute a competitor’s integration to you…. in fluent, authoritative prose
Left alone, that becomes what thousands of buyers believe.
Give Letaido your ground truth: your docs, your pricing page, your feature list, etc. Then get it to fact-check the AI assistants by pulling every AI answer about you from Brand Radar, check each factual claim against your source of truth, and flag the fabrications.
For each one, it traces the cited source that likely seeded it, so you know whether to fix a page, request a correction, or feed a better signal.
The result is a list of things AI is wrong about you, each with the receipt.
Build me an AI hallucination catcher. Inputs: my source-of-truth docs (pricing, feature list, product pages) and my brand. Use Ahrefs Brand Radar to pull every AI answer about me across platforms, extract each factual claim, and check it against my docs. Flag every claim that’s false or outdated, quote it, show the contradicting ground truth, and trace the cited source that most likely caused it. Output a fix list grouped by source so I can correct the worst offenders first.
Every AEO program eventually has to justify itself to someone who doesn’t care about citations, only outcomes. That report is tedious to assemble by hand and easy to put off, so it never gets made.
Get Letaido to write it.
On a schedule, it stitches together your AI share of voice and citation trends from Brand Radar, your search performance from Search Console, and your AI assistant referral traffic from Web Analytics into one client-ready monthly narrative.
KPI tiles, month-over-month movement, the prompts you gained and lost, and a short written summary a non-specialist can follow.
Build me a monthly AI-visibility report. On the 1st, pull my AI share of voice and citation trends from Ahrefs Brand Radar (ChatGPT, Perplexity, Gemini, Copilot, Google AI Overviews/AI Mode), my Search Console performance, and my AI-assistant referral traffic from Ahrefs Web Analytics. Compose KPI tiles, month-over-month deltas, prompts gained/lost, and a plain-English summary into one shareable report, and post a link to my #marketing Slack channel.
Final thoughts
AEO isn’t a project you finish. The answers move every week, and the defense is having something watching when you’re not.
That’s the real unlock here: Get Letaido to run all of these on a schedule and ping you when a competitor starts winning a prompt, when your pricing shows up wrong, and when sentiment slips.
If you’re an Ahrefs customer, you can try Letaido for free for one month.
Paste a starter prompt into a fresh workspace and you’ll have the first version running today. Then refine and customise it to your desired workflow. You’ll have your personal AEO tools in no time.
