Perplexity’s Shift to AI Agents Boosts Revenue 50% 

The company’s revenues have jumped 50% in the last month, with estimated annual recurring revenue (ARR) topping $450 million in March, the Financial Times (FT) reported Wednesday (April 8), citing figures seen by the news outlet.

As the report notes, the move marks a pivot from the chatbot-style search engine once viewed as a credible challenge to Google to AI agents that perform tasks on user’s behalf.

Perplexity has more than 100 million monthly active users from its search and agent tools, the FT said, citing company executives. This includes tens of thousands of enterprise clients, with the company making money via consumer and enterprise subscriptions, which can cost anywhere from $20 to $200 a month.

The FT points out that Perplexity’s growth is still dwarfed by that of other top AI startups. For example, coding company Cursor has seen its ARR reach $2 billion, up from less than $100 million in 2024. Anthropic reported an ARR of $19 billion at the end of February, while OpenAI said it generated $20 billion last year.

Writing about the rise of agentic AI last month, PYMNTS noted research showing that many consumers now use AI tools to assist with everyday duties like planning trips, researching purchases, organizing personal finances and learning about new subjects. 

“Instead of serving as occasional utilities, AI systems are becoming general-purpose assistants embedded across multiple aspects of daily decision-making,” that report said. “This shift matters because habitual use changes the starting point of digital activity.”

In other Perplexity news, the company this week introduced a new tax agent for Computer, its agentic AI designed to complete complex tasks with limited human supervision.

“The distinction between Computer and general-purpose AI chatbots is structural,” PYMNTS wrote. “Tools like ChatGPT or Gemini respond to tax questions based on training data that has a fixed cutoff date and no direct connection to current IRS materials.”

That report pointed to a test of four major AI chatbots conducted by TaxSlayer which found the tools miscalculated the refund or amount owed by an average of more than $2,000 in eight fictional tax scenarios, even when given the necessary forms.

Perplexity says it has addressed this by packaging tax knowledge as loadable modules built on its Agent Skills protocol. These modules are continuously updated and grounded in materials and regulations from the IRS, allowing the system to apply current rules.