AI BEACON
See what AI says about your brand,
and which page it read.
AI Beacon tracks how your brand appears across 7 AI engines + Google AI Overview, AI Mode & Search. When an answer names a source, you get the page it came from as well as the domain it sat on.
Rankings are measured. Answers are guesswork.
You can tell anyone where you rank for a keyword this morning. Ask which of your pages ChatGPT read before it recommended a competitor and the honest answer is usually a shrug and a screenshot.
A typical stack
- –Rankings in one tool, AI mentions in another
- –Answers captured by hand, one screenshot at a time
- –A domain credited, with no way to tell which page
- –No account of why a rival was named and you were not
With AI Beacon
- ✓Your prompts run across seven engines and Google AI surfaces
- ✓Presence in the answer read first, citations second
- ✓The exact page, thread or video the answer drew on
- ✓A breakdown of what each answer was built from
Presence comes first. Citations come second.
Whether a model names your brand inside the answer is the signal that matters most, because that is what the reader sees. The sources listed underneath tell you how the answer was built, which is what you act on.
AI Beacon holds both for every prompt you track, in one project, so a mention and the page behind it sit beside each other rather than in two separate tools.
Eight views of the same answer.
Each one has a page of its own. Start with whichever matches the question you are being asked.
Track the questions your buyers ask.
A prompt is the unit of visibility in AI search. Coverage is the set you have chosen to watch, and your standing on each one.
- Prompts you chooseAdd the questions your category asks, by market and by stage of the decision.
- Presence on each oneWhether your brand is named in the answer, and where it sits.
- Movement over timeThe same prompt, run again, so a change is visible rather than remembered.
Find the page behind every mention.
When an answer lists what it drew on, the useful detail is rarely the domain. It is the single page, thread or video that earned the place.
- The exact sourceA specific page URL, a Reddit thread or a YouTube video, named as the answer named it.
- Cited and consulted, kept apartA source quoted in the answer and a source the model read are two different things, and they are reported separately.
- Your pages, ranked by citationsWhich of your own URLs are doing the work, and which never get picked up.
Read how you are described, not only whether.
Two brands can both be named in the same answer and come out of it very differently. Sentiment scores the description itself.
- A score from 0 to 100One number per prompt, so a change in tone is as visible as a change in presence.
- Scored on the answer textWhat the model wrote about you, in that answer, on that prompt.
- Followed over timeA gradual slide shows up long before anyone notices it in a screenshot.
Watch the brands gaining ground on your prompts.
On any prompt you track, the model names someone. Seeing who, and how that changes, is how you notice a shift while it is still small.
- Share of the answers you trackWhich brands appear on your prompts, and how often.
- The sources behind their mentionsThe pages and threads doing the work for them.
- Changes worth a lookA brand that was absent last month and is named this month.
Take one prompt apart and see what the answer was built from.
Losing a prompt is not much use on its own. What helps is knowing which sources the answer used, and whether anything of yours was in the running.
- Every source the answer usedListed in full, with yours marked if it appears.
- Where your pages stoodWhether one of your URLs was read, quoted, or absent altogether.
- What the answer says about youThe description and the sentiment score for that prompt.
See what AI thinks your brand is.
Before a model can recommend you it has to have you filed as something. Entity mapping shows what it has filed, and what it attaches to your name.
- How your brand is classifiedThe category the answers place you in, in their own words.
- What gets attached to youProducts, people, markets and claims that travel with your name.
- Where it has gone wrongA wrong category or a stale fact, visible as text rather than guessed at.
Follow AI Overview, AI Mode and Search together.
Google is three results pages at once now. A page can hold position one and still be missing from the answer above it.
- AI OverviewWhether your brand is in the generated answer, and which pages it cites.
- AI ModeThe same question asked in the conversational surface.
- Beside classic SearchRead against where your pages rank, so the gap is obvious.
Join visibility up with what it brought in.
Being named is the start of the argument. Finishing it means showing the visits and the goals that followed.
- Visits from AI sourcesWeb Analytics records visits from AI tools, not only from the engines you track.
- Read against your promptsMovement in presence beside movement in traffic, in the same project.
- An honest account of the gapPlenty of AI visits arrive with nothing attached to say where they came from.
Seven engines, plus Google AI surfaces.
A prompt is run on each engine in its own right, because the same question produces different answers and different sources on each one.
SEORCE has observed 441.4M AI prompts: 33.9M from ChatGPT and 407.5M from Google AI Overviews, as of October 2026. That corpus covers those two engines only.
Four steps, from signup to answers.
The first session is the longest one, because choosing the right prompts takes some thought. After that you are reading, not setting up.
Add the site as a project
One project per site or market, so figures never get mixed between them.
Pricing is per project →Choose the prompts
The questions your buyers ask. Keyword Explorer is where you go looking for them.
Research prompts and keywords →Read the answers
Presence, sentiment and every source, per engine, with the same prompts run again over time.
Act on what they show
A gap becomes a page to write or a technical problem to fix, in the same project.
See what gets fixed →Eight capabilities, and the guide behind them.
Each page covers one capability in full, with the demo account on screen.
Seeing where you stand is the first of three jobs.
AI Beacon tells you where you are. The rest of SEORCE is what you do about it and how you show that it worked, reading the same project data throughout.
The SEORCE API covers AI Beacon and every other module, on Agency plans and above. See the API.
What is AI Beacon?
AI Beacon is the part of SEORCE that tracks how a brand appears when AI answers a question. It runs the prompts you choose across 7 AI engines plus Google AI Overview, AI Mode and Search, and reports whether your brand is named in the answer, how it is described on a score from 0 to 100, and which sources the answer drew on. Sources are resolved to the page, thread or video rather than the domain, and a source quoted in the answer is reported separately from one the model only read.
Key facts
- The engines are ChatGPT, Gemini, Perplexity, Copilot, Grok, DeepSeek and Llama, plus Google AI Overview, AI Mode and Search
- Brand presence inside the answer is the primary signal; citations are the second-order signal
- Citations name the exact page, Reddit thread or YouTube video, as well as the domain
- Cited and consulted sources are reported separately
- Sentiment is a score from 0 to 100, per prompt
- Each site or market is its own project, and the Visibility Score is scoped to one project
- AI Beacon data is on the SEORCE API, on Agency plans and above
- There is a free forever plan, with no card needed
Last reviewed 15 October 2026 by the SEORCE team.
Asked before people start.
Seven: ChatGPT, Gemini, Perplexity, Copilot, Grok, DeepSeek and Llama. Google AI Overview, AI Mode and Search are tracked alongside them.
Two things, in this order. Whether your brand is named inside the answer, which is what a reader sees, and which sources the answer was built from. Presence is the primary signal and citations are the second-order one.
The page. A citation is reported as the specific URL, Reddit thread or YouTube video the answer used, as well as the domain it sits on.
A cited source is one the answer quotes or credits. A consulted source is one the model read on the way to the answer without crediting it. They are counted separately, because the first earns you a mention and the second tells you what the model trusts.
As a number from 0 to 100 for each prompt, taken from how the answer describes your brand. Tracking it over time is the point: a description can soften long before presence changes.
No. The score is scoped to one project, so it is a measure of your own movement rather than a league table. For a view across brands, use the competitive intelligence page, which counts presence on the prompts you track.
No, and neither can anything else. Scores and gaps are a guide to where you stand and what to work on. The engines decide what to cite.
That depends on your plan. There is a free forever plan with no card needed, and paid plans are priced per project, never per seat. Anything beyond a plan's limits is priced custom.
No. Each site or market is added as its own project, so answers from different markets never get averaged together.
Web Analytics records visits from AI sources, and the attribution page reads those visits against movement in your prompts. Some of that traffic arrives with nothing attached to say where it came from, so the picture is directional rather than complete.
No. It shows the gap. Content Studio writes the page in your brand voice, and Technical Suite applies the technical fixes on your live site.
Yes, three ways. Build a dashboard and export it as Excel, PDF or Google Sheets, put a report on a schedule, or, on Agency plans and above, read it through the SEORCE API.
See what AI says about you today.
Add a site, pick ten prompts your buyers would ask, and read the answers you get back.