Launched BrandScope AI on Product Hunt today — it tracks how 6 AIs mention your brand
Just launched BrandScope AI on Product Hunt today 🚀
It tracks how 6 AI assistants (ChatGPT, Perplexity, Gemini, Qwen, GLM, Kimi) mention your brand in their answers — weekly reports on mention rate, competitor comparison, and blind spots.
Why I built it: I ran real founder brands through AI and found them completely invisible — one even got confused with a same-name company. AI answers are becoming the new discovery layer, and most founders have no idea what they say about their brand.
Free sample report on the site. Would love feedback from this community!
i like that you are treating visibility and identity as different problems. i would make the report split them into three buckets: the brand is not mentioned, it is mentioned but confused with another entity, or it is mentioned from a stale or wrong source. then each bucket can have its own next step. absence points to discoverability, confusion points to entity clarity, and stale citations point to source maintenance. that makes the weekly report easier to act on than a single mention percentage, even if the headline metric stays useful.
This is a genuinely better framing than what we have today — and honestly it maps to the data we're already collecting. The three buckets are: no mention (absence), mentioned-but-wrong-entity (confusion), mentioned-with-stale/wrong-source (source decay). We track absence and source-verified rate already, and we just hit the confusion bucket ourselves: Google's AI Overview now attributes "BrandScope AI" to a different company (wowai.ch) — our own product's problem, live.
We're going to restructure the weekly report around these three buckets, each with its own next step (discoverability / entity clarity / source maintenance), keeping the headline mention rate as the summary metric. Thanks for the push — it's the right shape.
In your IH title and post, the hook that grabs attention is tracking coverage across all 6 engines (specifically including Qwen, GLM, and Kimi).
On the live landing page subheadline, that gets collapsed down to just "OpenAI GPT Search, Perplexity or Google AI Overviews." You're dropping the non-Western/broader model coverage that actually differentiates the tool, and swapping Gemini (the model) for AI Overviews (the search feature).
If someone lands from your launch posts expecting the full 6-model tracker, the hero copy makes it look like a standard 3-tool wrapper.
Really good catch — and you're right that consistency matters here. Quick check on our side: the current landing hero leads with "AI Visibility (GEO) — ChatGPT · Perplexity · Gemini · Qwen · GLM · Kimi" and the metadata/OG copy all say "6 AI engines." So we did fix this at some point — but if you're still seeing the old "OpenAI GPT Search / Google AI Overviews" line, it might be a cached version, and I'd genuinely want to know where you hit it so we can hunt any straggler.
Your underlying point stands regardless: from launch posts, the first screen has to scream "we cover Qwen, GLM and Kimi too" — that's the differentiation, and it's a miss if any entry point collapses it to a generic wrapper.
caught it during the initial launch wave on the live root domain hero subtext, so it was almost certainly an edge cache purge delay on deployment. the updated line with the full 6-model list fixes the disconnect completely. glad it helped tighten things up. if you ever want a full second pair of eyes across the rest of the funnel as you scale traffic, happy to run through it.
The "confused with a same-name company" problem is more common than people think, we've run into look-alike name collisions with SocialPost.ai too, and it's usually not a monitoring problem, it's a consistency problem. The fix that's moved the needle for us is making sure our own site states positioning in the exact same wording everywhere, since LLMs tend to default to whichever source repeats a claim most consistently. Does your data show a founder's own site ever outranking third-party mentions as a citation source, or is that rare?
Great question — and your SocialPost.ai experience matches what we see. To answer directly: own-site citations are common, not rare — in our latest runs it ranged from 24% to 71% of all cited mentions across the 5 brands we track. Beryxa (an independent-evaluation service) is at 71% — its positioning statement on its own site is the single most-cited source. Plausible is at 48%.
So the "consistent wording on your own site" fix you found is directly supported by the data — when the brand's own domain is what the model cites most, the phrasing on that domain becomes the de-facto answer. The reverse is also worth noting: BrandScope itself sits at 24%, and the gap shows up as AI quoting directory listings or review sites instead of our own positioning.
The same-name-collision piece: we hit that with a brand whose name matches an unrelated company — the model cited the wrong company's domain entirely. Your consistency insight is exactly the right frame for it: it's not a monitoring problem, it's a "what does the most-repeated source say" problem.
the AI-citation angle from a thread I was in yesterday makes this doubly interesting, someone mentioned Reddit specifically gets cited 5-7% of the time regardless of question shape, and that they cite the same aged threads repeatedly rather than whatever's newest. so it's not just "does ChatGPT mention my brand" but "which specific threads is it pulling from and are they even accurate anymore"
does BrandScope show which source the AI is actually citing when it mentions a brand, or just that a mention happened? feels like the actionable part isn't the mention rate itself, it's knowing whether the AI is quoting a five-year-old outdated thread vs something current, since those would need completely different fixes
Great question — and honestly, you've put your finger on exactly where we're heading. Right now we show the mention, position, and context snippet (including the cited source URL when the model returns one) — so you can see what the AI is pulling from, not just that it happened.
The "stale source" problem you're describing is real: we've seen models quote old threads or confused same-name entities. Making source-age/accuracy a first-class signal in the weekly report is something we're actively building toward — the fix for "AI cites an outdated thread" is genuinely different from "AI doesn't know you exist."
If you'd like, I can run a sample report on your brand — happy to show what the source-level detail looks like today.
would genuinely love that, thank you. brand's StareBrain (starebrain.vercel.app), curious if there's anything out there yet given it's still pre-launch, might just come back empty which would be its own useful data point
and good to hear source-age is on the roadmap, that feels like the difference between the report being a vanity metric and an actual actionable tool
Ran it for StareBrain — report below. Even pre-launch, it's a genuinely useful data point: 80% of AI answers "mention" StareBrain, but almost none recommend it — Perplexity literally says "not yet available, cannot be recommended," Qwen only half-knows you (47%), and most mentions are name-drops, not endorsements.
That gap between mentioned and recommended is exactly what we're tracking — a pre-launch brand is the cleanest case: all awareness, zero decision impact.
Full breakdown (per-AI rates + per-question quotes with sources) below. When you launch, run it again and you'll see the mention→recommendation conversion start moving.
that's a genuinely surprising number for something this early honestly, curious how "mentioned" is being detected here, since 80% feels high for a brand with basically no indexed footprint yet. is it picking up genuine awareness, or is there a chance some of that is the model just generating a plausible-sounding "yes, StareBrain does X" answer without a real source behind it, which would show up as a mention but not really be one
either way this is useful, if it's real, that's a wild pre-launch signal worth understanding. if it's noise, that's also useful to know before I start citing an 80% number anywhere
Great question — and honestly, the right one to ask. You've caught exactly the thing we're still being careful about.
You're right that "mentioned" ≠ "verified." Here's what we actually measure today: we ask 6 AIs the same questions, and a mention is counted when the model names the brand in its answer. We do not yet claim that every mention has a real source behind it.
What the data shows (and why I'm glad you asked): across our latest runs, only Perplexity returns citations — 100% of its mentions have source links. The other five models (GPT, Gemini, Qwen, GLM, Kimi) return zero citations. So a brand's mention rate can be 80% while only ~30% of those mentions have a verifiable source. The rest is the model generating a plausible answer — which, for a pre-launch brand with little indexed footprint, is exactly the hallucination risk you're describing.
And to be fully transparent about the sample size: StareBrain's numbers come from 2 samples per question on the fast models (GPT/Gemini/Perplexity/GLM) and 1 on the slow ones (Qwen/Kimi) — so it's not a single snapshot, but it's also not a large-N measurement. Both the hallucination risk and the sampling noise point the same direction: treat the 80% as "the model is willing to say the name," not as awareness.
That's also why we're building this in layers. The next thing on our roadmap is exactly what you're poking at: separating "mentioned with a real citation" from "mentioned (possibly hallucinated)," and tracking source freshness — how old the cited sources are (the 5-year-old-thread problem you flagged earlier is the same disease). Until that ships, we label the report as single-snapshot and are transparent that mention rate ≠ verified visibility.
Genuinely useful push — if it's noise, better to know now than after someone builds a strategy on it.
this is exactly the kind of answer that makes me trust a tool more than a clean number would have, most founders would've just defended the 80% or quietly moved past the question. breaking down that only Perplexity actually cites, and being upfront about the 1-2 sample size, that's the transparency that actually makes a metrics product credible
appreciate you running it and being this straight about the limitations, genuinely more useful than if it had come back clean. will keep an eye on the source-verification layer once it ships, that's the version of this I'd actually build a strategy around
Appreciate that — the "clean number" temptation is real, and we decided early that for a metrics product, credibility > polish. If we can't show receipts, we say so.
Source-verification is already partially live: every mention now shows whether it has a citation, and models like Qwen/Kimi are flagged as "may be from memory." The freshness layer (source age) is next on the roadmap — that's the piece you'd build a strategy around, and it's coming.
Ping me anytime if you want to test it on a brand — happy to run a scan.
genuinely great to see the citation flag already live, that's fast turnaround from a comment thread to an actual feature. will take you up on testing again once StareBrain's launched and there's an actual before/after to look at, feels like the more meaningful test than running it again on the same pre-launch state
That's the right test, honestly. A pre-launch snapshot mostly measures noise — the before/after across an actual launch is where the "change" story lives, and it's exactly what a photo of the trend should show. Ping me the week StareBrain ships and we'll set it up.
The same-name company issue is particularly interesting. That sounds like a materially different problem from simply having low mention volume.
Exactly — same-name confusion is a different failure mode than low volume, and it needs a different fix. Glad that resonated.
If you’re open to continuing the conversation beyond the thread, what’s the best email to reach you at?
Appreciate you saying that! And yes — happy to continue beyond the thread. You can reach me at
support@brandscope.dev
.
Quick context on the same-name point: it's actually why we built the competitor-comparison angle into the report. If you've got a brand you'd like to check, happy to run a sample report on it — no card needed, just want to see what the AI answers look like for you.
This comment was deleted a month ago
Thanks! I’ve just sent it over.
Looking forward to hearing your thoughts whenever you have a chance.