In October 2025, my co-founders Upahar Sood, Gautham Seshadri, and I launched Snezzi, an AI visibility platform that helps brands understand how they show up when people ask ChatGPT, Perplexity, or Google AI what to buy, and improve it.
This isn’t a post about how AI search works. It’s about what building this company has taught me about building any company. Most of the lessons were about positioning, trust, and distribution, not technology.
Why I’m building this
For more than a decade I’ve grown products through organic channels: at Shoppr, at Whatfix, at Plivo. The playbook was always some version of the same thing. Understand what people search for, build content and authority, earn the traffic.
Then I started noticing something in our own data. A few visitors were arriving from ChatGPT. Not many at first, but they converted better than almost any other source, because by the time an AI recommends you, the trust work is already done.
A few years ago, “Do we rank on Google?” was a question every brand could answer. Now the question is “Does ChatGPT recommend us?” and almost nobody can. I got tired of waiting for someone else to solve that properly.
Lesson 1: Lead with the problem, not the category
I made a mistake early on. I kept describing what we did as “AI SEO.”
Every conversation derailed into “So you do backlinks and keyword research?” and I’d spend twenty minutes explaining why this was different.
The conversations changed when we stopped leading with a category and started leading with a question:
When someone asks ChatGPT for the best CRM for startups, does your product show up?
That lands immediately, because every founder and marketing lead has used AI to research products themselves. They already know the answer matters.
I’m still learning how to explain it to different audiences. SaaS teams get it fastest. Ecommerce brands want to see data first. Agencies want to know how they can offer it themselves. But the principle holds: building a new category is messy, and the way through is to describe the buyer’s problem in the buyer’s words.
Lesson 2: Ignore the debate, show the work
Right after we launched, the internet was busy arguing about whether this whole space was real, hype, or just SEO in disguise.
We decided not to argue the theory. Every new category goes through a skeptical phase. Some people dismiss it. Others spot the opportunity early and build. The only argument that convinces anyone is results, so we focused on doing the work, being honest about what’s still uncertain, and sharing what we learned along the way.
That also meant setting honest expectations. Organic growth has always taken time and consistent effort. AI doesn’t change that. It only changes where a brand needs to show up. Under-promise, over-deliver.
Lesson 3: Give away the right piece
A $5 Chrome extension put Snezzi inside 60 browsers in 30 days. 97% of those installs came from the US, and we’re building from India.
The extension audits any webpage for technical SEO and AI visibility. It’s free and needs no signup.
A lot of advice treats Chrome extensions as a backlink tactic. The backlink is nice. But the real value was who installed it. Someone installing our extension isn’t casually reading an article. They’re actively auditing a page for the exact problems we help with. By the time they reach us, they’ve already seen how we think.
And because AI has made software so cheap to build, building the extension wasn’t the hard part. Choosing which piece of our product to give away was. Give away too little and nobody cares. Give away the wrong thing and you attract people who’ll never buy. Give away too much and there’s no reason to talk to you.
Lesson 4: Earn trust with skeptical buyers
Many of the buyers we talk to have been burned before by agencies and retainers that produced reports but no results. Their skepticism is earned.
Recently a founder told me most of the proposals he’d seen sounded the same. We showed him the specific buyer questions we track, where his business currently appeared in AI answers, and exactly what we’d work on. His remaining concern was whether it would deliver.
So we proposed a pilot. He could follow the work and results in our platform, and part of our fee would only be payable once we hit an agreed KPI. He said we were probably the only provider willing to do that. He decided to work with us.
Putting part of your fee at risk isn’t right for every deal. But when trust is the real objection, sharing the risk is often the clearest way to answer it.
What I’d tell another founder
- Describe the problem before the category. If people keep misunderstanding what you do, your words are the problem, not them.
- Don’t wait for the debate to settle. Do the work and share the receipts.
- Give away something people use at the exact moment they feel the problem.
- Answer skepticism with shared risk, not louder claims.
We’re still early, and I’ll keep sharing what we learn, including what doesn’t work.