This week, Perplexity is introducing a new agentic tool for its subscribers. Dubbed Perplexity Computer, the company describes it as a system that “unifies every current AI capability into a single system.” Specifically, it functions as a computer user agent capable of independently executing complex workflows by leveraging 19 different AI models, even generating subagents to tackle specific tasks. The tool is available immediately, but exclusively on the platform's top subscription plan, Perplexity Max, which costs $200 per month. It operates fully in the cloud, a design the company suggests may alleviate some of the security concerns associated with other agentic tools.

Last week, Perplexity executives held a background briefing to discuss this product and outline their annual agenda. The event was originally scheduled to include a live demonstration, but the company canceled that portion just hours beforehand due to discovered flaws in the product.

This launch marks a significant evolution for Perplexity. The company first gained attention early in the AI boom by integrating advanced AI models into user-friendly interfaces, most notably its search-engine-like answer service. It continued this trajectory last summer with the launch of the Comet web browser. According to one executive, competitors like Google have since adapted their own products to resemble Perplexity's offerings—a trend seen as both a compliment and a competitive threat.

The company is actively adapting to a changing market landscape. After being one of the first AI firms to experiment with advertising, Perplexity discontinued that business late last year, stating last week that ads had compromised user trust in the accuracy of its answers. However, Perplexity's user base, which numbers in the tens of millions, is substantially smaller than that of OpenAI, which reports 800 million weekly users and began testing ads in ChatGPT this year.

In response, Perplexity executives now describe a strategic pivot toward a more specialized, boutique user base. They aim to serve professionals making significant, high-stakes decisions, which they refer to as “GDP-moving decisions.” The briefing emphasized a focus on growing enterprise subscriptions, particularly for deep research applications. “You don’t hear us talk about MAUs ever, because we’re not actually on a mission to get as many users as possible,” one unnamed executive noted.

To support this direction, Perplexity recently released a new benchmark for complex research tasks called Draco, where its own deep research product reportedly outperforms competitors like Gemini. The company also highlighted that it no longer relies on other companies' APIs for its web index and now operates its own AI-optimized search API.

A core part of Perplexity's strategy involves packaging cutting-edge AI models within a consumer-friendly experience. The company argues there is inherent value in orchestrating multiple third-party large language models (LLMs) to deliver the most cost-effective and accurate responses. “Multi-model is the future,” asserted one executive, contending that models are becoming more specialized, not commoditized.

Internal data supports this approach, showing users frequently switch between models for different tasks. For instance, in December 2025, queries for visual outputs were most often sent to Gemini Flash, software engineering tasks were handled by Claude Sonnet 4.5, and medical research was directed to GPT-5.1.

Perplexity's software can automatically select the optimal model for a given task, such as choosing one LLM for coding and another for drafting marketing copy. Executives also cited the use of their own modified, open-source Chinese-built LLMs to answer queries more affordably—a technique that drew criticism last year for not being transparently disclosed to customers. The company believes that when applied openly, this method can efficiently optimize LLM usage.

Another feature, called Model Council, allows users to query multiple models simultaneously. While the economics of providing multiple queries under flat subscription rates present challenges, executives expressed confidence. They claim to have high margins on user fees and, without costly infrastructure projects, believe they can stay competitive by intelligently allocating computational resources to the best model for each purpose.

Looking ahead, Perplexity plans to release its Comet browser for iOS next month. The company will also host a developers' conference named Ask on March 11 in San Francisco to encourage third-party use of its API. This commercial focus is becoming more pronounced internally; one executive mentioned now reviewing daily revenue metrics instead of query volumes.

This shift toward monetization has not gone unnoticed by users. The Perplexity subreddit has seen frequent complaints about new rate limits being applied to both free and subscription tiers. Company executives, however, dismissed these concerns during the briefing. “Any discussions on the free tier being made worse or rate-limited is completely false,” one executive stated.