People used to search. Now they ask.
Spinlap builds Radar, an AI search optimization platform. It tells a company whether AI engines name them when their buyers ask — and then helps make sure they do.
What changed
Ten years ago you typed “best CRM for small business” and got ten blue links. You clicked one, and a website got a visitor. Today you ask ChatGPT, or Perplexity, or you read Google’s AI Overview at the top of the page — and the answer names two or three companies without you clicking anything at all.
So the old goal, be number one on Google, is worth less than it was. The new one is be one of the companies the answer names.
Here is the part that made us start building. Any business can check its Google ranking in five seconds with a dozen free tools. But ask an owner whether ChatGPT mentions them when a customer asks for a recommendation in their category, and they have no idea. They cannot check. There was no tool in their hand that told them. That is a real, painful, unanswered question — and unanswered questions are where products come from.
One more thing changed, and it matters for what we chose not to build. In 2023, “AI writes your blog posts” was a product. Now everyone has that, and search engines actively discount sites publishing generic AI content in bulk. So a business cannot be built on writing. It has to be built on knowing what works.
What we actually do
We measure
We take the questions a company’s buyers would actually put to an AI and ask them — to several engines, repeatedly, over weeks. We record who was named, who was linked, and who was left out. Repeatedly, because these answers are not stable.
We explain
A number on its own changes nothing. So we look inside the answer: which passage the engine quoted, whose page it came from, and whether what it says about you is even true — the wrong price, a feature you dropped, the wrong audience.
We fix
Engines do not quote whole pages, they lift one passage. So the unit of work is the passage: not “write a post about X”, but write the paragraph an engine would quote instead of a competitor’s.
We prove it
Then we keep measuring, so there is a before and an after. “You published this on the 3rd; on the 12th Perplexity started citing you for this question” is the sentence the whole product exists to be able to say.
Who it is for
The user is anyone who cares whether AI recommends them — a SaaS company, a clinic, a law firm, an online shop. The buyer we build for is an agency, and that is a deliberate choice. An agency already sells “we handle your visibility” to ten or thirty or fifty clients, and those clients have started asking about AI. Giving one agency an answer is worth more than convincing thirty individual owners.
That is also why the audit is free and needs no login. Most people who run it will never buy, and that is fine: it costs us little, it is the cheapest way for anyone to get a straight answer about their own brand, and it is how we get found at all.
What we deliberately do not do
- We are not building a better AI writer. That market is finished. Writing is a step inside the product, not the product.
- We do not sell raw data. We could dump AI answers into a spreadsheet and sell it. It is a one-time sale with bad margins, and the engines’ own terms forbid reselling their output — doing it would cost us the access the product depends on. We sell conclusions, not the raw material.
- We do not use one customer’s data for another without permission. The system does get better by learning across results, and where that happens it is written into our terms in plain words rather than buried.
- We do not scrape the AI products. Official APIs only. Scraping breaks weekly and violates their terms.
- We do not chase every engine. A handful, covered properly, beats a long logo wall. Coverage is not the hard part; depth is.
The company
Spinlap is a software product company, founded in 2024 and based in Chandigarh, India. Radar is what we build. We are not a services or consulting studio, and we are unrelated to similarly-named organisations elsewhere.
If you want the plainest possible version of what we are betting on, it is this: any single feature here can be copied in a few weeks. What cannot be copied is the record — which sentences engines quoted, from whom, and what happened after someone acted on it. That only accumulates by being collected as it happens.
Questions are welcome on our contact page, and the blog is where we write up what we learn as we go.
Start with your own number
The fastest way to understand what Radar does is to point it at your own domain. The audit asks the major answer engines a real question set and shows you who they cite instead of you.