iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
When Luxcerta founder Chuan Peng (Gary) asked grounded Gemini what his company was, it repeatedly treated “Luxcerta” as a likely typo for LuxCarta—and did not mention, cite, or retrieve Luxcerta’s website. Peng reports that Google Search’s AI Overview described the company correctly on the same day. After he made several web identity changes, the Overview continued to give an accurate description about a week later. The account shows how different AI surfaces can represent the same brand differently; it does not prove that any one website change caused the later result.
What Peng measured in Gemini
Peng says he tested grounded Gemini with five identity questions, including “What is Luxcerta?” and “What does luxcerta.com do?” He asked each question twice in a fresh conversation, for ten answers in total. He counted mentions of the company name, citations of luxcerta.com, and retrieval of the site using plain string matching.
| Measure | Peng’s reported baseline |
|---|---|
| Answers mentioning Luxcerta | 0 of 10 |
| Answers citing luxcerta.com | 0 of 10 |
| Answers retrieving the site | 0 of 10 |
Instead, he says Gemini associated the name with LuxCarta and suggested the domain might be parked or phishing. These are Peng’s measurements from a small, self-reported test, not results independently reproduced by a third party. Peng’s account on DEV Community was published September 17, 2026.
Why Gemini and Google Search differed
On the same day as the Gemini test, Peng says Google Search’s AI Overview described Luxcerta as an independent studio focused on GEO monitoring. That is a comparison between two specific products and their retrieval systems—not evidence that one is generally more accurate than the other. For a company checking how it appears in AI answers, the surface matters: an answer in a chat product is not interchangeable with an answer in a search results feature.
#1 Best Overall
What Peng changed
Peng reports making several changes intended to present a consistent, crawlable description of Luxcerta:
- Added Organization structured data in JSON-LD to the homepage.
- Made the homepage meta description clearer.
- Updated a GitHub profile page to repeat the company’s description and explicitly distinguish Luxcerta from LuxCarta.
- Submitted a sitemap and requested a recrawl through Search Console.
About a week later, he says Google’s AI Overview gave a stable, accurate one-sentence description and no longer showed the earlier LuxCarta association on that surface. The timing is an observed sequence, not a causal test: the changes were bundled, there was no control group, and the account does not isolate which action—if any—made the difference. Nor does a week of consistent answers establish that the result will persist.
How to check what AI says about your company
Peng’s method suggests a practical way to make brand monitoring more interpretable: define what counts before looking at the answers, and keep observations tied to the product and conditions under which they were obtained.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →- Choose identity questions. Include both a direct name query, such as “What is [company]?”, and a domain query, such as “What does [domain] do?”
- Record the product and settings. Note the AI product, whether web grounding or search features are enabled, the date, and whether each prompt is asked in a fresh conversation.
- Repeat prompts consistently. Reusing the same wording and session procedure makes changes across observations easier to interpret.
- Set explicit counting rules. Track company mentions, links or citations to the site, and whether the site appears to have been retrieved as separate outcomes. A mention alone does not show that the site was cited.
- Keep the raw answers. Save responses alongside the counts so a later review can distinguish a genuine citation from a name mention or an incorrect attribution.
- Repeat over time and across surfaces. AI answers can vary, so a single result—or a short period of consistency—should not be treated as a permanent change.
What the case does—and does not—show about structured data
Organization JSON-LD was one of Peng’s changes, but the account does not establish that structured data alone makes an AI system describe a company correctly. The clearer meta description, disambiguating GitHub profile, and sitemap and recrawl actions happened in the same intervention window. The defensible takeaway is narrower: Peng tried to make the company’s identity more consistent across several crawlable pages, and later observed a better answer in Google’s AI Overview.
Rank #3
A separate example of inconsistent AI recommendations
Peng also summarizes a survey of dental implant clinics in greater Taipei. He says he asked 15 real-user questions in three fresh conversations each, producing 45 answers per platform. In his account, an answer sometimes cited one clinic’s website while recommending another, and recommendation lists across ChatGPT, Claude, and Gemini barely overlapped. The article reports these observations but does not provide the underlying dataset or enough detail to verify the survey independently, so they should be read as his reported example rather than a general measure of recommendation accuracy.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
