How to Build an AI Search Business That Competes

TL;DR

Build an AI business by solving a distinct customer problem, shipping quickly, and improving retention as the company grows. Perplexity differentiated itself through reliable search-based answers when competing assistants were less capable, stayed efficient by focusing on product quality instead of training proprietary models, and attracted investors with visible usage, a strong team, and clear written strategy.

Transcript

give me three examples of businesses which are going to benefit from the rapid adoption of AI So software development healthcare financial consulting What is the most ridiculous search query you have read Either this person is trying to rob a bank Ended up asking I want my face protected I still want to be able to breathe but it should cover my nos... Read More

Key Insights

• Perplexity is an AI metric that measures how well something is understood, and its domain initially cost $120 for two years. The name was selected when the company lacked funding and had not yet decided exactly what search product or consumer experience it would build.
• Perplexity's early strategy was to create search experiences that had not previously been possible. The founders knew they wanted to work in search, but they were not initially planning around a billion-user product or following a fully defined consumer application roadmap.
• Jeff Bezos approved his first investment after reviewing a memo and a personalized product demonstration. The demo imagined Bezos speaking with a voice system about subjects including Star Trek and Blue Origin, showing how a tailored presentation can make an emerging product concrete.
• Written strategy documents are tools for sharpening a founder's thinking, according to the discussion. Preparing memos and six-page updates requires significant effort, but the process creates clarity and allows investors to focus meetings on specific assumptions, uncertainties, and strategic decisions.
• Customer retention is a central measure of healthy growth. Bezos advised Perplexity to keep acquiring users without chasing unreasonable growth, prevent the product from becoming a leaky bucket, and ensure retention improves as the user base expands rather than merely remaining flat.
• Perplexity differentiated itself by offering a search service that the founder says was not reliably available through competing assistants at the time. Visible usage, rapid product releases, efficient spending, credible investors, and a clear product distinction made the startup more compelling.
• Perplexity's product philosophy treats the underlying model as an implementation concern rather than the customer's concern. The company spent its limited early funding on continually improving the user-facing product instead of training proprietary models, reflecting a strong focus on delivered customer experience.
• Disruption often occurs when an established company must protect its margins while a newcomer uses better technology to deliver more value. The founder identifies Google as having the most to lose if Perplexity succeeds, although Bezos argued that Perplexity does not require Google's failure to succeed.

Questions & Answers

Q: How did Perplexity choose its company name?

Perplexity was chosen when the founders knew they wanted to work on search and create previously unavailable experiences, but they did not yet have a precise consumer product plan. The word refers to an AI metric that measures how well something is understood. The domain perplexity.ai was available for $120 for two years, which made it practical for an unfunded company.

Q: How could Perplexity improve its difficult consumer name?

The founder openly accepts that Perplexity is difficult to pronounce and is not an ideal name for a consumer company. One possible solution is to keep Perplexity as the corporate name while giving the product a shorter, simpler name. He also suggests that greater funding and usage could eventually allow the company to acquire rights to a concise four-letter or five-letter name.

Q: How did Jeff Bezos become an investor in Perplexity?

Perplexity approached Bezos through a shared contact and was asked to submit a written memo. The founder also created a personalized demonstration in which an imagined version of Bezos spoke with a voice system about Star Trek and Blue Origin. Bezos reviewed the materials and approved the initial investment without first meeting the founder, according to the interview.

Q: Why are written memos useful when presenting an AI startup?

A written memo forces the founder to organize the company's product, strategy, assumptions, and unresolved questions with precision. The Perplexity founder says producing memos and six-page documents takes substantial effort but improves clarity. It also enables an informed investor to study the material beforehand and use meeting time to discuss specific sections instead of repeating an introductory presentation.

Q: What growth advice did Jeff Bezos give Perplexity?

Bezos advised Perplexity to prioritize customer experience and keep users after attracting them. He warned against becoming a leaky bucket, where acquisition rises but customers do not remain. His guidance was to maintain growth without targeting something unreasonable and to make sure retention improves as the company expands, rather than allowing retention to stay flat or deteriorate.

Q: Why did Perplexity stand out from other AI products?

Perplexity stood out because its search service delivered an experience that the founder says competing assistants did not offer reliably at that time. ChatGPT was not browsing reliably, while Google Bard, Microsoft Copilot, and Bing Chat were described as less capable. This gave Perplexity a clear period in which it could attract users as a differentiated product rather than another similar assistant.

Q: How did Perplexity use its early funding efficiently?

Perplexity raised relatively little money before the investment round discussed in the interview and avoided spending that money on training its own models. Instead, the team repeatedly invested in improving the product customers used. Its underlying principle was that the model was not the customer's concern. Customers cared about receiving the expected result, regardless of the internal cost or technical arrangement.

Q: How can an AI search startup compete with a large company?

An AI search startup can compete by delivering a clearly better customer experience, releasing improvements quickly, spending efficiently, and retaining users as growth increases. The interview also presents disruption as a situation where an established company must protect existing margins while a newcomer exploits better technology. Perplexity sees Google as having the most to lose, but its success does not require Google's failure.

Summary & Key Takeaways

• Perplexity began with a broad interest in creating new search experiences rather than a detailed plan for a massive consumer application. Its name came from an AI metric and an inexpensive available domain. The founder acknowledges that the name is difficult for consumers and says the company and product could eventually use different names.

• Jeff Bezos initially approved an investment after receiving a written memo and a personalized demonstration involving voice interaction, Star Trek, and Blue Origin. Later, he requested a six-page strategy update and discussed it in detail. His advice emphasized customer experience, sustainable growth, stronger retention, and avoiding a business that behaves like a leaky bucket.

• Perplexity gained attention because its search service offered a differentiated experience when other assistants did not browse reliably or perform as well. The company used limited funding efficiently, avoided spending heavily on training its own models, and concentrated on delivering useful results. Its founder identifies Google as the competitor with the most to lose.