A Complete Guide to Using Peec AI (with Actionable Takeaways)

Peec AI is an AI visibility tool where you (as the user) track prompts for your brand (or client) across different AI assistants (like ChatGPT, AI Mode, AI Overviews, Gemini, or Claude) and record metrics like visibility (mentions) in resulting chats as part of a GEO strategy.

As a former employee of Peec AI, I have a comprehensive understanding of the tool and how to take action from its data.

In this guide, I’ll walk you through Peec AI from top to bottom, explaining different areas of the tool and teaching you how to take action to improve your visibility as part of a GEO strategy.

One note before we begin: the information and screenshots in this guide are up to date as of July 20, 2026.

Update: I published a video to accompany this article. Feel free to check that out, as well!

To understand Peec AI, the best place to start is with the Prompts section.

Prompts are the basis for the data you see in Peec AI (or any prompt-tracking tool; a lot of the advice in this guide also applies to tools like Profound, of which I’m also a power user).

Peec AI works by sending out your designated prompts to different AI assistants, recording the resulting chats, and then analyzing them for visibility metrics (like mentions) as well as sources or citations.

We’ll get into all of that later, but first, it’s important to set up your prompts correctly from the start:

Ideally, your prompt set should represent the full buyer journeys of your target personas across your key lines of business.

One mistake I’ve seen brands and agencies make is having an incomplete prompt setup, which means your visibility won’t be truly representative of your target personas’ buyer journeys through AI assistants.

No tool currently can show you what real users are prompting their AI assistants, so to get prompt ideas for your Peec AI setup you can:

• Talk to customers (interviews, sales calls, or CX data)
• Mine Reddit threads
• Check your keyword data (like from GSC)

If you have a current setup of prompts, I’ve found it helpful to export those prompts, upload them to Claude (or another AI assistant), and ask where the gaps are in your target personas’ buyer journeys for your key areas of business.

You can also use Claude to brainstorm prompt ideas from scratch, but the more real user data you start with, the more directionally accurate your prompts are likely to be.

Now we’ll discuss Topics, which is how you organize your prompts.

Topics are like folders for your prompts in Peec AI. You typically want your topics to align with your core lines of business. This makes it easier to filter your visibility metrics by business area (which we’ll discuss later).

Each topic should have around 5-10 prompts in it to start, but the more prompts you add, the more variables you’re likely to cover.

Prompts within topics should also share common traits, like having the same competitors or content strategy.

Every prompt can have one topic, but it can have multiple tags.

Tags allow you to slice your data within Peec AI.

Peec AI automatically assigns tags for branded or non-branded as well as various intents (funnel stages).

You can also add tags for target personas (like I did for Expert and Novice) or any other criteria that you think would be helpful for interpreting your data: geographic locations, products/SKUs, comparison prompts, etc.

When you have a representative prompt set organized by Topics and Tags, it makes it easier to filter your data, which is important for taking action (which we’ll discuss later).

Speaking of filtering, let’s hop to the Overview tab.

The Overview tab shows your visibility score (based on mentions in tracked prompts) against tracked competitors. You can also see your share of voice, sentiment, and average position.

We’ll dive into the “All Domains” section in a moment, but that’s where we’ll start taking action.

The first step with the Overview tab is to filter your data by Timeframe, Tags, Model, and Topic (and Location, if you’re tracking multiple countries).

We want to filter our data because AI visibility can vary between models (based on the sources involved).

There’s nothing wrong with looking at your visibility score globally, but if you want to take action, filtering is the first step.

Based on the above filters, we’d be looking at AI visibility data for the last 14 days for chats triggered by Informational prompts in the “Holistic SEO / User-First SEO” Topic in AI Overviews.

Ready to take action? Let’s hop into the “All Domains” section.

You can visit the Domains section using either the “All Domains” button on the Overview tab or the Domains side button under Sources:

For this exercise, I’ll use the following filters with the Generative Engine Optimization topic (where I have low visibility and would like to improve):

Once you have your data filtered, the next step is to identify the sources that are influencing the chats associated with your tracked prompts.

You can start by looking at Domain Types:

AI assistants can answer user queries from their training data alone, but if additional information is needed, the AI assistants will ground their answers (using RAG) with external sources.

Those sources (the Domain Types listed above) are your ticket to a GEO strategy, because influencing the sources an AI assistant references is how you can potentially influence its answer in your favor (ideally to recommend your brand and cite you).

Each Domain Type invites its own strategy:

• Corporate: take inspiration for your own content or look for partnership opportunities.
• UGC: take inspiration for your own social or video channels or find creators to partner with.
• Reference: ensure your brand is represented in these sources, either directly or indirectly.
• Editorial: do digital PR, reach out to journalists for coverage, or explore paid advertorials.
• Other: explore the types of sources there and how you can influence them with owned or earned media.
• Competitor: take inspiration for your own content by identifying gaps (competitive analysis).
• Institutional: explore partnership opportunities.

We can explore the sources associated with a Domain Type by filtering for it:

Then, you can identify a source domain (in this case, we’ll go with Profound’s website) and view all of the URLs that are being retrieved as sources for the chats associated with the prompts we’re tracking:

Similarly, we can go to the URLs section and view all URLs being used as sources in the chats associated with our tracked prompts.

Here is a look at the top URLs used as sources for the chats associated with our informational Generative Engine Optimization prompts:

Based on this high-level data, we can see that a YouTube video would be a good asset to create, as well as a how-to-style article.

But owned content and socials is only part of the game. What third parties have to say about your brand is equally (or perhaps more) important. That means finding opportunities to get mentions.

We can use the Gap Analysis tool in Peec AI to find all URLs where tracked competitors are being mentioned but we’re not.

In the above screenshot, we see that those URLs mention at least three tracked competitor brands but not our brand, indicating that we could reach out to them to potentially get mentioned in those or future articles.

Note: the “competitors” in my setup are brands I admire for their content, not actual competitors.

But the effectiveness of a GEO content strategy may be limited if relevant AI bots can’t access the content. Luckily, there’s a section of the tool for that, as well.

This section is called Crawlability, and it shows us whether our robots.txt file is blocking any AI training, search, or user-query bots:

You’ll note that my robots.txt partially blocks bots; that’s because I disallow internal search pages and a few other pages from crawling.

It’s important to investigate whether AI bots can reach your site’s content.

On that note, check if you’re loading content with client-side rendered JavaScript. I can recall sites that weren’t getting AI visibility, and when I explored their technical setups, they relied on CSR JavaScript instead of server-side rendered JavaScript, which likely contributed.

I’ll show you another cool part of the tool that I find handy for reporting purposes: Insights.

Within Insights is a table called the Performance Matrix. You can customize the X and Y axes to view the data you want, such as seeing how you’re performing for different Topics in different models you’re tracking:

Speaking of content strategy, I almost forgot a key part of the tool …

Let’s talk about Fanouts.

I’ve written about fanout queries on my blog before, but in case you want a quick refresher: when some AI assistants perform web retrieval, they may decompose a prompt into several related searches (fanout queries). These fanouts can cover different aspects of the user’s intent and help the system identify relevant passages and sources.

Peec AI allows you to see fanout queries from ChatGPT (and Perplexity, but I don’t track that model):

These queries can be used to optimize your content directly, but a better application is to use the fanout queries as inspiration for the intents your content should address.

The more comprehensive your content is in terms of its relevance to fanout queries, the greater your likelihood of being retrieved as a source for them.

Here’s a neat trick that I’ve seen others write about as well: export your fanout queries from Peec AI (assuming the file size is manageable), upload them to Claude, and ask it to extract general themes that you can use for content ideas.

Key takeaways:

• Peec AI is an AI visibility tool where you designate prompts to be tracked and the tool records the resulting chats in AI assistants (like ChatGPT, AI Mode, AI Overviews, Gemini, or Claude) and analyzes them for metrics like visibility (mentions), citations or sources, sentiment, and average position.
• Your visibility score is only the beginning. You can take action on the data by first filtering by Timeframe, Tag, Model, and Topic and then analyzing the sources being retrieved in the chats for the prompts you’re tracking.
• Owned media is a big part of the game (like taking inspiration from Corporate or Competitor content), but earned media can be especially powerful (like striking up partnerships with Corporate sites or reaching out to Editorial domains for coverage; and don’t forget about UGC, especially Reddit, which I plan to do another post about specifically).

This is a quickly evolving topic, so I’ll definitely return to this article to update its information or improve the formatting over time.

And, of course, if you want help with your prompt-tracking setup (whether it’s in Peec AI, Profound, or another tool) or you’d like a hand creating a GEO strategy from your AI visibility data, I’m available as an independent SEO/GEO consultant.

Until next time, enjoy the vibes:

Thanks for reading. Happy optimizing!

Related posts