The New Discovery Channel Creators Can’t See (And Why It Matters More Than Your Follower Count)
If you run social accounts for a living, you already know the sinking feeling of watching organic reach get squeezed, algorithm changes gut your best-performing format, and platforms quietly deprioritize the link in your bio. You’ve probably spent the last two years optimizing for watch time, saves, and shares. But here’s the uncomfortable shift happening underneath all of that: a growing slice of your audience isn’t discovering brands through feeds at all anymore. They’re asking ChatGPT, Gemini, Perplexity, and Claude “what’s the best [thing] for [problem]” — and getting a shortlist of names that may or may not include yours. That’s the problem MentionBird is trying to make visible, and it’s worth paying attention to even if you never touch the product, because the mechanics behind it are about to reshape how creators and social teams think about distribution.
What MentionBird actually does
MentionBird is a tool from maker Salil Panikkaveettil — a founder who says he spent a decade helping brands get found on Google before building this. The pitch, as stated on the Product Hunt launch page, is straightforward: you enter your company and URL, and the tool finds the buying questions that matter in your category, tracks daily which brands ChatGPT, Gemini, Perplexity, and Claude name in their answers, and identifies which sources those answers are built on. Then an “Answer Agent” turns that into a ranked action plan of which sources to get into and what to publish or fix.
The team also shipped a rebuilt Answer Agent and a new YouTube Answer Agent, and they’re offering the first three months of the Starter plan for $19 with code PH19 for the Product Hunt community. Pricing beyond that promo isn’t disclosed in the thread. If you want to dig into the discussion itself, the thread lives here.
The core insight Panikkaveettil states plainly: “AI answers aren’t built from your rankings. They’re built from the sources AI trusts: review sites, listicles, forums, publications. If you’re not in those, you’re not in the answer.” That sentence is the whole thesis, and it’s the part social media operators should internalize regardless of whether they buy the tool.
Why this is a social media problem, not just an SEO problem
Here’s where I’d push back on the framing that this is purely a search-marketing play. In my experience running content across Instagram, TikTok, YouTube, and LinkedIn, the sources that AI models lean on overlap heavily with the content social teams already produce — but the format and the distribution logic are completely different.
Think about how a model like Perplexity or ChatGPT with browsing assembles an answer to “best CRM for a small agency.” It’s not pulling from your Instagram grid. It’s pulling from comparison articles, Reddit threads, YouTube reviews, and listicles. That means your YouTube channel, your founder’s LinkedIn posts, your podcast transcripts, and your community threads are all potential citation sources — if they’re structured in a way that’s extractable.
Why TikTok creators should care more than LinkedIn ones
This is counterintuitive, so let me explain. If you’re a TikTok creator whose entire value proposition is entertainment and reach, AI recommendation tracking is largely irrelevant to you. Nobody asks ChatGPT “what’s the funniest dance creator.” But if you’re a creator who sells a product, a course, a template pack, or a service — or if you run social for a brand that does — then the AI answer layer is quietly becoming a top-of-funnel channel you can’t measure with your current analytics stack.
LinkedIn creators, ironically, may care less than they think, because LinkedIn content rarely surfaces as a citation source for AI answers. It’s gated, it’s noisy, and models tend to favor open, indexable, text-heavy sources. The platforms that matter most for AI visibility right now are the ones with strong public text: YouTube (via transcripts), Reddit, review sites like G2 and Trustpilot, and your own blog. That’s a useful mental model for where to invest repurposing effort.
The measurement gap nobody’s talking about
When I scheduled 30 posts across five platforms last month for a client, I could tell you exactly what each one did: impressions, engagement rate, saves, click-throughs with UTM parameters. What I could not tell you is whether any of that content contributed to the brand showing up when someone asked an AI assistant for a recommendation in that category. There’s no UTM for “the model mentioned us.” That’s the gap MentionBird and its competitors are trying to fill, and it’s a real one — even if the methodology is inherently fuzzy.
How it stacks up against the incumbents
The most useful exchange in the entire launch thread came when a commenter asked how MentionBird differs from Profound and Scrunch. Panikkaveettil’s answer is refreshingly honest: “We’re in the same category as Profound and Scrunch, and the core tracking (prompts, mentions, citations, competitors) is similar.” He then names three differentiators: showing the actual web searches the AI runs before answering, uncapped MCP access so you can pull data into Claude, ChatGPT, or your own agents, and pricing built for smaller teams rather than enterprise budgets.
That’s a credible positioning. Profound has positioned itself toward larger marketing organizations, and Scrunch has built a following among SEO-adjacent teams. If you’re a lean social team or an indie founder, the “affordable tier of a category that’s currently enterprise-priced” pitch is the same playbook that worked for tools like Buffer against enterprise social suites, or Metricool against Hootsuite. My take: the MCP access angle is the sleeper feature here. If you’re already running any kind of AI-assisted workflow — using Claude or custom GPTs to draft content or analyze performance — being able to pipe your AI-visibility data into those same tools means you can build your own dashboards and agents without waiting for the vendor to ship a feature.
Where the honest limitations live
The maker is admirably clear about what he can’t promise. When asked whether following all the Answer Agent’s recommendations guarantees results, he says: “No, and anyone who guarantees AI visibility should make you suspicious, because nobody controls what the models say.” That’s the correct answer, and it’s worth repeating because the entire category is riddled with overclaiming. What MentionBird promises is that recommendations come from real data — what engines answer for your prompts, which brands they recommend, which sources they cite — and that daily tracking lets you see whether changes move the needle.
There’s a second honest admission worth flagging. When asked how the tool determines which prompts potential customers would ask, Panikkaveettil concedes: “Nobody can see what people actually type into ChatGPT, so every tool in this space is making an informed estimate.” MentionBird starts from real search keyword data and converts high-intent keywords into likely AI questions, and you can edit or add your own. That’s a reasonable methodology, but it’s an estimate, not ground truth. Anyone selling you “exact AI query data” is bluffing.
What social teams can borrow from this, tool or not
Even if you never sign up, there are three operational moves worth stealing from the way this product frames the problem.
First, start treating AI answers as a distribution channel you audit quarterly. Pick ten questions a real buyer in your category would ask an AI assistant. Ask them across ChatGPT, Gemini, Perplexity, and Claude. Screenshot the answers. Note which brands appear and which sources get cited. Do this manually, once a quarter, and you’ll learn more about your category’s AI visibility than any dashboard will tell you in month one.
Second, look at your content through a “citation-worthiness” lens. A 60-second TikTok rarely gets cited by a language model. A well-structured YouTube video with a clear transcript, a Reddit-style community answer, a comparison blog post, or a detailed review on a third-party site does. If your repurposing workflow currently turns one long-form asset into fifteen short vertical clips, consider whether one of those outputs should instead be a text-heavy, extractable piece — the kind of thing an AI model would actually want to quote.
Third, connect this to your existing analytics discipline. You already track UTMs, engagement rate, and watch time. Add a column to your monthly report for “AI mentions observed” — manual, imperfect, but better than nothing. Over two or three quarters, you’ll start to see whether your content strategy is moving the needle on the channel that’s quietly eating the top of your funnel.
Where the math breaks
I’d be cautious about over-indexing on any single AI-visibility tool right now, for a structural reason: the models change constantly, the retrieval sources shift, and what works today may not work in six months. A tool that tracks daily is only as good as its prompt set, and if your prompt set doesn’t match what real buyers ask, you’re optimizing for a phantom. This is why the “edit your own prompts” feature matters more than it sounds — it’s the difference between tracking your category and tracking someone’s guess about your category.
Where I think it falls short
A few open questions the launch thread doesn’t resolve. The Starter plan’s exact feature limits beyond the promo price aren’t disclosed. How many prompts you can track, how many competitors, how deep the historical data goes — all unstated. The YouTube Answer Agent is mentioned as recently shipped but there’s no detail on what it actually does differently from the core product. And the biggest unknown: how MentionBird handles the inherent volatility of AI answers. If ChatGPT gives a different shortlist on Tuesday than it did on Monday for the same prompt, does the tool surface that variance, or does it smooth it into a misleading trend line? That’s the kind of thing I’d want to test before trusting any “visibility score.”
There’s also a category-level caveat. This space — AI visibility tracking — is crowded and moving fast. Profound, Scrunch, and a rotating cast of newer entrants are all chasing the same insight. Some will consolidate, some will pivot, some will get acquired. Betting your workflow on any one of them this early is a calculated risk, not a safe default. My take: use one to learn the mechanics, but keep your prompt sets and audit notes in a doc you own, so you’re not locked into a vendor’s interpretation of your category.
What I’d watch / test next
This week, I’d do three things. First, run the manual AI audit I described — ten buyer questions, four models, screenshot everything. It takes an afternoon and it will tell you more about your actual AI visibility than any onboarding flow. Second, if you’re curious about MentionBird specifically, use the PH19 code to test the Starter tier for three months and see whether its tracked prompts match the questions you’d actually want to win. Compare its competitor list against your own — the maker claims Starter auto-surfaces competitors that AI actually mentions, which is a genuinely useful feature if it holds up, especially for local or niche businesses that don’t know who they’re really competing against. Third, add an “AI mentions” line to your next monthly report, even if it’s just a manual count. The operators who start measuring this channel now, while it’s still fuzzy and unglamorous, will be the ones who understand it when it stops being optional.
