I'll be honest – I never expected to be tweaking our content for the likes of ChatGPT and Perplexity. Yet here I am, a founder who spent years appeasing Google's algorithm, now learning to please the AI overlords. 😅

The results speak for themselves:

• FlashDocs became the #1 result inside AI chat for our category
• We dominated searches for "best APIs to automate slide decks"
• Achieved this with $0 spent on ads

How? We embraced a new playbook called Generative Engine Optimization (GEO) – basically SEO for AI answers.

In this guide, I'll share exactly how we did it, in five tactical sections. It's candid, a bit witty, and 100% from our own trenches. If you're a founder or growth lead trying to make your brand unmissable in ChatGPT or Perplexity answers, read on. (And yes, GEO is a thing – think of it as making sure AI models cite you when your buyers ask them questions.)

🧠 1. How We Identified the Exact Prompts Our Buyers Use

The first step in GEO was surprisingly human: figuring out what our customers were actually asking the AI. We couldn’t optimize for AI answers until we knew the questions. Here’s how we nailed down the exact prompts our buyers use:

• We put ourselves in the buyer’s shoes (literally). I imagined I was a dev lead searching for a tool like FlashDocs. Would I type something vague like “slide API”? Nope – I’d ask in plain English: “What are the best APIs to automate slide decks?” or “Who can generate PowerPoint slides via API?” We listed out these natural-language questions, not just keywords.
• Dug into our search data and customer chats. We peeked at Google Search Console to see which phrases led people to our site. Then we transformed those into questions. For example, if “automate Google Slides API” was a common query, we turned it into “How can I automate Google Slides with an API?” – exactly how someone might ask ChatGPT. We also combed through our customer emails and chats for the wording they used. (You’d be amazed – your users have already written your best prompts for you in their feedback and questions.)
• Asked the AI itself. In a move that felt like peeking at the quiz answers, we literally asked ChatGPT and Perplexity what questions people might ask about our category. This meta approach gave us a few prompt variations we hadn’t thought of. (Yes, I asked an AI how to optimize for AI – we’ve come full circle.)
• Benchmarking with competitors. We tried prompts like “Compare [Our Product] vs [Big Competitor]” and “Tell me about [Our Company]”. Early on, the answers were humbling – ChatGPT had never heard of us, or mixed us up with Flash something else. But this exercise showed which questions were common and who was being mentioned. It gave us a target: the exact queries we needed to dominate.

By the end of this process, we had a shortlist of high-intent prompts our buyers were likely using. Think of it as our AI-age keyword list, except it was full-sentence questions. This became the foundation for everything else – you can’t optimize for what you haven’t identified. (And if you’re wondering, yes, “best slide deck API” was at the very top of our list – along with a few egocentric ones like “Is there an API for Google Slides?” and “FlashDocs vs Google Slides API”.) Knowing these, we were ready to make sure we were the answer.