In a sea of AI models now flooding the market, Perplexity stands out as the underdog, making waves among tech giants like GPT and Gemini. Despite arriving late to the party, this scrappy RAG AI—short for Retrieval-Augmented Generation—has redefined how we interact with AI by blending web searches with generative responses. With answers enriched by live data and backed by real-time sources, Perplexity is taking search engines to task, possibly heralding the end of “paid-for” search results as we know them.

AI Professional – Fiona Passantino, early November, 2024

Fetch!

Amidst the flood of new models that come and go, new systems that seem to ship weekly, one small model seems unique, catching the eye of the larger players like Amazon and Nvidia. Perplexity is a (relatively) young, innovative player that came late to the game but was able to change how we define and use AI despite having no “baked-in” user base (such as Google) and no hardware to roll it out on (such as Apple or Microsoft).

Perplexity is a RAG AI, which stands for “Retrieval-Augmented Generation”. It’s a Large Language Model that combines web searching with generative capabilities. The responses it delivers are fed partially by its own training data but also by the live web, making it more relevant thanks to information found in the wild, more knowledgeable with respect to your query and more accurate in its answers. Importantly, it provides live links so we can do our essential Human fact-checking on the spot. In many ways, RAG AI combines the best features of generative tools and traditional search engines, delivering quick, relevant answers with sources and images pulled from the live web.

What will this mean for the future of traditional search? Given the way Google has grown into more of an ad factory and a pusher of “sponsored” (paid for) results, it would not be surprising if the Rise of the RAG may well lead to the Death of Search as we know it today.

How does it work? The Perplexity interface looks like a standard AI input space – minimalistic and friendly, not much in-window functionality and no agents. You input a prompt that takes the form of a question: What’s the best way to explore Corsica by moped? How many kilos of chocolate do Swedes eat per year? The RAG fetch-doggie first combs through its historical data for relevant information that does not require a live web response, then trawls the internet to update any missing “holes” in its set. Then, it generates a response like a ChatGPT might. The big difference is, the aggregated AI response is backed by links to specific websites you can check and verify; an essential part of the AI-Human workflow.

Perplexity went big, fast. Launched in 2022, Perplexity AI suddenly reached a $1 billion valuation after just two years of service. Still free for the user – for now – it’s still growing quickly among consumers and business users. It’s not clear how it will pay for itself in the long run. Either ads embedded in its responses, like Google, or with a subscription, or both. For now, this model and most others are being kept afloat with buckets of investor cash, which currently seems limitless. But at a certain point, the wolves howling for ROI will overpower the visionaries and all LLMs must learn to carry their own weight.

Perplexity AI is still running on the endless drip of investor cash. Sooner or later, the free ride will end, and this RAG will have to figure out how to pay the bills—whether through ads, subscriptions, or both. For now, we’re all here for the ride.

---

Need help with AI Integration?

Reach out to me for advice – I have a few nice tricks up my sleeve to help guide you on your way, as well as a few “insiders’ links” I can share to get you that free trial version you need to get started.

---