LLMagnet

The discovery layer your social strategy is quietly missing

If you run social for a brand, a creator business, or a portfolio of client accounts, you already know the uncomfortable truth of the last eighteen months: the click is dying, and the answer is eating it. People ask ChatGPT, Claude, Gemini, and Perplexity for recommendations the way they used to type queries into Google, and your carefully scheduled Instagram carousel or YouTube tutorial is increasingly just source material for an answer you never see. That shift is why a WordPress plugin called LLMagnet caught my attention — not because social teams should rush out and install it, but because it names the problem the rest of our stack has been ignoring: we have dashboards for what happens after the click, and almost nothing for what happens before it. If you’re a social operator, that pre-click blind spot is now your problem too.

What LLMagnet actually does — and what it doesn’t

Let me strip the launch copy down to mechanics, because the framing matters more than the feature list. LLMagnet is a WordPress plugin that adds an “AI visibility layer” to a site. Per the maker’s launch post, it does roughly five things: tracks real AI bot activity and shows which content those bots access; measures an “AI Visibility Score”; generates and maintains llms.txt and llms-full.txt; improves structured data and machine-readable content; and connects the site to AI clients like ChatGPT, Claude, and Cursor via an MCP connector.

The llms.txt piece is the one I’d flag first for anyone who’s been sleeping on it. It’s a proposed convention — think of it as a robots.txt for the LLM era — that gives AI systems a curated, markdown-friendly map of your most important pages instead of forcing them to parse your whole DOM. If you’ve ever watched a crawler burn your server budget re-reading paginated archives, you understand why this matters.

Here’s the part I want to be precise about, because the launch thread got muddy on it. When a commenter asked whether the tool shows “views without clicks,” the makers were refreshingly careful. Ben Cohen clarified that LLMagnet tracks crawler requests, which can happen with no human involved — and that you cannot infer someone saw your brand recommended just because a crawler visited. That’s an honest distinction most AI-visibility startups blur. I’d bet a lot of competitors are quietly conflating “we saw GPTBot” with “ChatGPT recommends you,” and it’s the kind of thing that gets a tool distrusted the moment a marketer checks the math.

Where the visibility score actually comes from

The other thing worth unpacking is the score. A commenter asked point-blank how it’s calculated, and the answer is more modest than the marketing implies. The Overview Visibility Score is, in the maker’s words, “a readiness checklist for AI access/understanding — not ‘does ChatGPT recommend you.’” The dashboard breaks it into crawl frequency, bot quality, traffic volume, page health, and content quality. So it’s closer to a Lighthouse audit for AI-readability than a ranking. That’s a meaningful reframe: a high score means your site is legible to machines, not that machines are citing you. Two very different things, and only one of them pays rent.

Why this matters more to social operators than it looks

Stay with me here, because on the surface this is a WordPress/SEO story, not a social one. But the plumbing connects. In my own workflow, the content that performs best on social is almost never only social — it’s the blog post, the recipe, the case study, the landing page that the caption points at. When I’ve scheduled 30 posts across five platforms in a month, the entire point of the UTM discipline is to route that attention somewhere I control. If that somewhere is a WordPress site, and AI systems are increasingly the ones summarizing it to the next person, then your social distribution and your AI discoverability are the same funnel viewed from opposite ends.

Think about a food creator. A recipe page gets crawled, gets cited in an AI answer to “easy high-protein dinners,” and — maybe — earns a click from someone who wanted the photography and the voice, not just the ingredient list. The maker made exactly this point in the thread, arguing that the AI mention alone “doesn’t pay the bills” and that the returning reader is still the prize. That’s the correct frame, and it’s the same frame any social manager should apply: AI visibility is a top-of-funnel signal, not a revenue line.

Why TikTok creators should care more than LinkedIn ones

Here’s a distinction the launch thread didn’t make but operators should. The value of AI-crawler visibility scales with how reference-shaped your content is. A TikTok creator whose value is the performance — the face, the edit, the sound — has less to gain from being cited by an LLM, because the LLM can’t reproduce a dance. A LinkedIn ghostwriter or a B2B content team, by contrast, is publishing exactly the kind of structured, factual, quotable material that AI systems vacuum up and paraphrase. So if you’re in the second camp, this category deserves your attention now; if you’re in the first, watch it but don’t panic.

How it stacks up against the incumbents you already pay for

The most useful question from the whole launch came from a commenter named Daniel, who asked how this differs from Google Analytics and Google Search Console — the tools every site owner already has. It’s the right question, and the answer the makers gave is defensible: GA measures people after they click, while LLMagnet sits “earlier in the chain,” showing which AI crawlers hit you and which URLs they reached. The two complement rather than replace each other.

But I want to be honest about the competitive field, because the launch copy doesn’t name anyone and I think creators should. The AI-visibility category now includes tools like Profound, Peec AI, and Otterly.ai, most of which focus on tracking brand mentions inside AI answers across platforms. LLMagnet is playing a different, narrower game: it’s a WordPress-native crawler-and-readiness tool, not a cross-platform answer-monitoring suite. If what you actually want is “show me every time ChatGPT mentions my brand,” this isn’t that product — at least not yet. If what you want is “tell me which bots are hitting my site and whether my pages are structured for them,” that’s squarely the target.

The MCP connector is the sleeper feature

The piece I’d watch hardest is the MCP connector for ChatGPT, Claude, and Cursor. MCP — Model Context Protocol — is the emerging standard for letting AI clients talk to external tools and data. Wiring your WordPress site into an AI client means you could, in principle, query your own site’s content and structure conversationally. For a content team, that’s a genuinely interesting workflow: “which of my service pages have weak headings?” asked in plain English. It’s early, and the launch thread didn’t demo it, so I’m flagging it as promise rather than proof.

Where my judgment says it falls short

Now the part the makers won’t write, and that I owe you as someone who’s tested a lot of half-baked marketing SaaS.

First, the score is a proxy, and proxies get gamed. The moment “AI Visibility Score” becomes a KPI, someone will stuff their pages with machine-readable fluff and watch the number climb without any real citation gain. The makers are clear the score isn’t a recommendation metric — good — but the market will treat it like one anyway. That’s a trust risk baked into the category, not just this product.

Second, the WordPress lock-in is real. If your site runs on Webflow, Squarespace, Framer, or a headless stack, you’re out of scope entirely. The maker confirmed the plugin needs a WordPress user with plugin-install permissions — usually Admin — and no FTP or pixel setup. That’s clean, but it’s also a wall. Not disclosed: whether a non-WordPress path is on the roadmap.

Third, “AI visibility” is a moving target. The AI answer engines change their crawling and citation behavior constantly, the way Google’s algorithm shifts. A readiness checklist is only as good as its checks, and the checks are only as good as the platforms’ current behavior. I’d want to see how often the scoring criteria get updated before I’d trust it as a north-star metric.

Fourth, the honest limitation the makers themselves surfaced: this is an added layer, and adding layers is exactly what overworked social and content teams don’t need. As one commenter put it, if a food blogger has to become a “GEO expert” to get value, you’ve just given her another job. The makers’ answer — that they’re pushing toward specific, actionable recommendations (“this page needs attention, here’s what’s missing versus the pages being cited”) — is the right direction, but the launch thread describes it as roadmap, not shipped. The recommendations are the product. The dashboard is just the receipt.

What I’d watch / test next

If you run content for a WordPress-based brand or client, here’s what I’d actually do this week — no hype, just steps.

1. Install it on one low-stakes site and watch the crawler log for seven days. The first question the maker built for is “which agents read my site” — so answer it for yourself. You’ll likely be surprised by which bots show up and which pages they touch.
2. Audit your llms.txt situation. Whether or not you use LLMagnet, check whether your site has one and whether it points at the pages that actually matter. If it doesn’t exist, that’s a five-minute fix with outsized leverage.
3. Cross-check against Google Search Console. If your AI-crawler traffic and your organic search traffic point at the same pages, your content strategy is aligned. If they diverge, that’s a signal worth chasing.
4. Don’t rebuild your social calendar around this. Treat AI visibility as a measurement layer, not a content strategy. Your captions, hooks, and watch-time still do the heavy lifting.
5. Pressure-test the recommendations when they ship. The whole value proposition lives or dies on whether the tool tells you what to change, not just what’s wrong. Until that’s live, calibrate your expectations accordingly.

The bigger picture: discovery is fragmenting, and the operators who win the next two years will be the ones who can see across both the social surface and the AI substrate underneath it. LLMagnet is one early, WordPress-shaped attempt at the second half. Whether it becomes infrastructure or a footnote depends on whether it ships the part that tells you what to do — because a dashboard alone, as the makers themselves admitted, “wouldn’t be enough.”