The problem with asking an AI a factual question has never been that the answer is usually wrong. It is that you cannot tell when it is. Perplexity's answer to that is simple and effective: every sentence comes with a footnote you can click.
How it changed my research
Over two months I used it for pricing comparisons, market figures, regulatory questions and background reading. The pattern that emerged: Perplexity is fastest at questions with a definite answer scattered across several pages — the kind that would take four tabs and ten minutes on Google.
The citations are doing real work. I caught two answers built on a low-quality source within seconds of scanning the links, something no citation-free assistant would have allowed.
Pro Search and Spaces
Pro Search slows down and does more: multiple queries, clarifying questions, deeper synthesis. It is noticeably better on comparative questions.
- Good: 'compare pricing models of X, Y and Z' with sources for each
- Good: extracting figures from a report you upload
- Weaker: highly technical niche topics where the best material is not indexed
- Spaces keep a project's threads and files together — useful for a long research task
Where Google still wins
Local results, shopping, anything where you want to see the range of opinion rather than a synthesis, and any query where the source you want is a specific site you already trust. Perplexity is a summariser; sometimes you want the raw list.
Pro versus free
Free is genuinely usable and caps Pro Searches per day. The upgrade buys volume, model choice and file handling. Two weeks of real use will tell you whether you hit the ceiling — if you do, you will hit it every day.
Verdict
Perplexity has not replaced my search engine or my AI assistant; it sits between them and it has taken over the part of the job I liked least. For anyone who has to be right in public, the citations alone justify the subscription.