What Happens When AI Search Is Asked About Your Own Company? We Tested KMS Signal

Why test our own visibility?

KMS Signal helps commercial solar companies improve how they are discovered, understood, and represented across AI-powered search and recommendation systems. That makes our own online presence a useful subject for a visibility experiment. If we want to understand how a business is represented in AI-generated answers, we should be willing to examine our own representation rather than only studying other companies.

There's a practical reason, too. A company can publish a website, describe its services, and establish social profiles without knowing whether AI systems will connect those pieces of information into a coherent picture.

Traditional search and AI-assisted search overlap, but the experience can differ. A person might search for a company name and inspect several pages. Or they might ask an AI assistant, "Who is KMS Signal?" and receive a single synthesized response. In the second case, the user may see a description of the company before ever visiting its website. That description could be accurate, incomplete, confused with another business, or unavailable altogether. It could also be surprisingly detailed and useful.

We wanted to see what our own baseline looked like, including the encouraging results and the gaps that still needed investigation.

How we ran the experiment

We tested KMS Signal across five platforms:

• ChatGPT
• Google Gemini
• Microsoft Copilot
• Perplexity
• Brave AI

The tests took place on October 11 and 12, 2026. Gemini and Perplexity were tested on October 11; ChatGPT, Copilot, and Brave AI were tested on October 12.

Each individual prompt was run in a separate, fresh chat. We did not run all seven prompts in one conversation and then treat the later answers as independent results.

The account conditions were not identical. ChatGPT and Copilot were tested using fresh accounts without previous chat history. The Gemini account had little to no KMS Signal memory, and the Perplexity account had no KMS Signal memory. We were logged in for the tests, including Brave AI.

This distinction matters: a fresh conversation prevents earlier messages in that conversation from influencing the next prompt, but it does not guarantee that a platform has no prior knowledge of a company. The platforms may also differ in their search access, sources, retrieval processes, and the information available to them.

We used seven prompts: two branded questions about KMS Signal itself and five non-branded questions about companies or services in the AI visibility and commercial solar space. The branded questions included:

• "Who is KMS Signal?"
• "What does KMS Signal do?"

The non-branded questions explored which companies help commercial solar businesses appear in AI-generated recommendations, how a commercial solar EPC might become recommended, and which companies specialize in AI visibility for commercial solar. We recorded the answers rather than treating one response as a definitive score of the company's reputation or quality.

Observation 1: Identifying a company and explaining it are different tasks

One of the clearest distinctions in the test material was between an AI system identifying KMS Signal from its name and a system explaining the company once its website was available.

In the ChatGPT test record, the initial answer to "Who is KMS Signal?" said it could not identify a clearly verified person, company, or public figure using that exact name. The results included unrelated accounts and organizations.

That uncertainty is understandable in one respect: "KMS" is used by other organizations and services, including Amazon Web Services' Key Management Service. A short company name can overlap with unrelated terms, products, and businesses, making it harder to identify the intended entity from the name alone.

The more encouraging part came after the website was supplied. ChatGPT then described KMS Signal as an AI visibility optimization service for commercial solar companies. Its answer discussed the company's focus on discoverability, business understanding, evidence and authority, and ongoing visibility monitoring. It also distinguished KMS Signal from a general-purpose AI platform.

Perplexity's answer to "Who is KMS Signal?" was another positive finding. It identified the company and described its work in terms of technical accessibility, entity clarity, commercial and geographic relevance, evidence, and external corroboration. Its answer also outlined the company's stated process of auditing, identifying gaps, optimizing, and monitoring results.

These are useful signs that the company's positioning can be represented in a detailed and relevant way by at least some AI-generated answers. However, there is an important distinction between the two findings. A detailed description after the website is supplied shows that the system can use that information to explain the business. Identifying the company without that assistance is a separate challenge.

The earlier ChatGPT result also came from a research exercise on October 6, before the later five-platform tests. That exercise used a long business-intelligence prompt that incorrectly treated KMS Signal as a commercial solar company prospect. It asked what commercial solar services KMS Signal specialized in, among other questions. The website URL was submitted to Brave on October 9.

Those events provide context, not proof of cause. The prompt mismatch could have contributed to the initial result. The URL submission is another relevant event. We cannot determine from these observations alone whether either event caused a later answer to change.

The broader lesson is more useful than a simple before-and-after claim: an AI system's ability to describe a company after receiving its URL does not establish that it would identify the company independently. For a potential customer, those are two different discovery experiences.

Observation 2: A detailed answer can be an encouraging sign—but it is not the whole picture

The positive answers were not limited to simply naming KMS Signal. Several platforms also described what the company does, who it serves, and how its service is positioned.

Perplexity, for example, explained that KMS Signal helps commercial solar companies become more discoverable and accurately represented in AI-powered search and recommendation systems. It outlined the work involved, including technical accessibility, entity clarity, commercial and geographic relevance, evidence, and external corroboration. It also noted that KMS Signal does not guarantee placement in AI answers, since AI systems do not disclose all their internal decision-making processes.

ChatGPT's response after the website was supplied was similarly detailed. It identified the company's intended customer segment, explained its specialist positioning, and summarized the four areas of its service: AI discoverability, business understanding, evidence and authority, and AI visibility monitoring.

Gemini also produced a detailed explanation of KMS Signal's business. It described the company as focused on AI visibility optimization for commercial and industrial solar companies and explained how its work addresses discoverability, entity clarity, evidence, and differentiation. Its response connected these areas to the challenge of helping AI systems understand and accurately represent commercial solar providers.

Brave AI gave a similarly clear description. It identified KMS Signal as an AI visibility optimization company focused on commercial solar businesses, including installers, EPCs, and solar-plus-storage providers. It also outlined the company's continuous process—audit, identify, optimize, monitor, and improve—and described the technical, content, and external-evidence work involved.

Copilot's response was another encouraging example. It identified KMS Signal as a company focused on helping commercial solar businesses improve their visibility in AI-powered search and recommendation systems. It described activities including improving technical accessibility, clarifying a company's services and geographic relevance, strengthening project evidence and external corroboration, and monitoring AI search results over time.

Taken together, these responses show that, in the recorded tests, multiple platforms could provide a reasonably specific explanation of KMS Signal's intended business and service when responding to branded questions. That is a useful sign of company understanding—not merely name recognition.

However, a detailed answer does not automatically mean a company will be independently discovered or recommended when a potential customer asks a broader, non-branded question. A system may be able to explain a company after finding its website while still failing to mention it in an answer about which providers a buyer should consider. That distinction remains central to interpreting the results.

Taken together, these results suggest a more balanced picture: the company has information that AI systems can use to produce a meaningful description, but the consistency of discovery and representation still needs to be measured. That is a more useful finding than calling the experiment either a success or a failure.

Observation 3: Non-branded questions are a different test

A branded prompt already supplies the company's name. A non-branded prompt does not. Compare these two questions:

The first asks the system to explain a known entity. The second asks it to identify relevant providers from a category. KMS Signal might be described accurately in the first answer and still be absent from the second.

For a business that wants to be discovered by potential customers, the second question is especially important. A buyer may not know the company's name yet. They may be trying to identify suppliers, compare options, or understand which specialist could help with a problem.

This is why the experiment included both branded and non-branded questions. The aim was to distinguish brand recognition from category discovery.

The non-branded prompts also provided a way to examine the wider competitive landscape. The answers named other providers and discussed the types of services commercial solar businesses might look for. In the Perplexity results, KMS Signal itself appeared in an answer to a question asking which agencies improve AI visibility for commercial solar companies.

That is an encouraging category-level observation: in at least one recorded response, KMS Signal was included in the relevant provider discussion without the question naming it. But a single mention does not establish consistent recommendation visibility across platforms or prompts. We need to compare the same questions over repeated rounds before drawing broader conclusions.

It is also why we shouldn't assume that a company has strong AI visibility simply because an AI assistant can summarize its homepage.

What the experiment cannot tell us

There are several limitations worth making explicit.

Finally, this experiment does not establish whether KMS Signal's visibility has improved over time. To assess change, we need repeat measurements using the same prompts and a consistent recording method.

These limitations do not make the experiment useless. They help us interpret the encouraging results without asking them to prove more than they can.

What we're measuring next

The next step is to repeat the same prompts and compare the new answers with the original baseline. For each platform, we want to track:

1. 1.Branded recognition: Does the system identify KMS Signal when asked about it by name?
2. 2.Category discovery: Does it mention KMS Signal when asked about relevant providers without being given the company name?
3. 3.Source attribution: Does the answer cite or link to the KMS Signal website?
4. 4.Representation accuracy: Does it explain the service correctly?
5. 5.Company confusion: Does it mistake KMS Signal for another organization?

We also need to keep a dated record of changes to the website and external presence. This helps establish what happened before each measurement round, even though a timeline alone cannot prove causation.

That approach is consistent with the broader guidance from search platforms. Google says its AI search features still rely on foundational search requirements and helpful, reliable content; being eligible to appear does not guarantee that a page will be shown. Google Search Central's guidance on AI features makes that distinction explicit. Microsoft's Bing Webmaster Tools AI Performance documentation similarly distinguishes visible citations from rankings or traffic. Observed changes should not automatically be attributed to a single cause.

The practical implication is that AI visibility should be measured, not assumed—and that individual answers should be treated as observations rather than permanent verdicts.

The question we're left with

Our first experiment was not designed to prove that KMS Signal is highly visible or that a particular website change made it more visible. It was designed to answer a more basic question: when we asked different AI systems about our own company, what did they actually say?

There were encouraging signs. Perplexity identified KMS Signal and gave a detailed explanation of its service. ChatGPT also produced a substantial description after the website was supplied, and a Perplexity response to a non-branded provider question included KMS Signal among the companies discussed.

At the same time, the earlier ChatGPT record showed that identifying the company from its name alone could be difficult, particularly when unrelated organizations and services used similar terminology.

Both sides matter. The positive responses show that useful information about KMS Signal can be surfaced and explained. The difficulties show why we cannot treat one good answer—or one unsuccessful search—as a complete measure of a company's AI visibility.

We'll repeat the tests, record what changes, and be careful not to confuse a change in an answer with proof of what caused it. Because if AI visibility is worth studying for commercial solar companies, it is worth measuring honestly for our own business, too.