Perplexity's Model-Agnostic Flexibility
Perplexity at $51M ARR
Perplexity’s edge is that it can swap the engine and redesign the dashboard at the same time. Because it is not tied to owning a search index like Google or a single frontier model like ChatGPT, it can mix Google and Bing retrieval with multiple LLMs, then ship whatever interface best fits a query, from finance research to team workspaces to internal file search. That makes new surfaces a product decision, not a platform rewrite.
-
Perplexity started by answering long, messy knowledge work questions by pulling live web results into GPT generated summaries with citations, then moved upmarket with paid prosumer and enterprise tiers. That model agnostic stack let it optimize for the query and workflow, instead of being locked to one model or one ad driven search page.
-
The contrast with enterprise search is concrete. Glean sells broad internal search across SaaS apps, while Hebbia goes deeper into high stakes workflows like diligence, contract analysis, memo writing, and pitchbook generation. Perplexity can test both directions quickly, from general search interfaces like Spaces to enterprise features like Internal File Search.
-
This flexibility matters because interfaces are becoming the moat. Glean also uses a multi model approach and has expanded from search into an internal agent builder, which shows where the market is going. The winner is less likely to be the company with one best model, and more likely to be the company that turns retrieval, reasoning, and workflow into the fastest improving product surface.
From here, Perplexity is likely to keep fragmenting search into specialized experiences where answers lead directly into work. As model quality converges, the advantage shifts to whoever can turn the same underlying models and search inputs into the most useful interface for each job, and ship those interfaces faster than incumbents can reorganize around them.