The Prompting Company is a platform that gets your product cited on ChatGPT, Perplexity, Gemini, and more. As more customers turn to AI for recommendations, you need to show up in those answers to stay visible. They’re trusted by world-class companies like Rippling, Zapier, and many others, including us.
Behind the curtain, The Prompting company uses AI-powered browser automations to analyze real user-facing answers across these LLMs, so it can see exactly what customers see. But they didn’t start out that way.
Problem: When an API isn’t actually enough
Prior to Kernel, their first attempt at building this product was to use the LLM’s APIs directly. Intuitively, that seemed like the right move. If there’s an API available, why not use it? But it quickly became clear to Albert and the team that the API results didn’t match what real users actually saw. In the UI, real responses include web results and citations that can heavily influence a buyer’s decision.
“We consistently saw the API behave differently from the UI because of its inherent limitations, and that just didn’t allow us to produce accurate enough results for our customers.”
Albert Purnama, Co-founder & CTO, The Prompting Company
They needed to see exactly what customers see in their browser, not just what the APIs return, and they needed to do this at large scale.
Solution: Selecting Kernel over Browserbase
Every customer request could take one to two minutes end to end: you have to open the browser, open ChatGPT, and then wait for the task to finish. To analyze thousands of prompts at once, The Prompting Company runs them in large batch jobs and needs to run many of these tasks in parallel to keep total processing time reasonable. For that reason, The Prompting Company’s key requirement wasn’t shaving seconds off a single run, but ensuring high throughput.
Albert first tried to build on top of Browserbase in the early days of the product, but ran into scaling issues once they started doing serious analytics at scale. In contrast, their initial tests with Kernel ran without a hiccup.
They also needed a way to manage credentials so they could safely access and operate across many of their paid accounts on ChatGPT and other LLMs. That’s when they started trying our profiles, which let them persist and reuse cookies, logins, and session data across their fleet of browsers.
“We evaluated other options, but really Kernel was a no-brainer. We wanted to make sure we didn’t cut any corners.”
Albert Purnama, Co-founder & CTO, The Prompting Company
Results: 50,000+ browsers every month
Today, The Prompting Company runs its AI-powered browser automations on Kernel, spinning up 50,000+ browsers, each collecting an LLM’s response to a particular prompt, and giving their customers fast, accurate data on how their product ranks on ChatGPT and others.
On top of the scalability, they get hands-on support from our team.
“The Kernel team is super-responsive on Slack. We haven’t had to wait more than 5 minutes for a response. The difference was night and day compared to dealing with other providers.”
Albert Purnama, Co-founder & CTO, The Prompting Company
If you’re currently limited by an API, let us help. We’ll get you around those limitations.
