Perplexity AI Inc., a startup that spent three years trying to dismantle Google’s search hegemony, is now betting its $20 billion valuation on a pivot from answering questions to doing the work.
The San Francisco-based company founded by Chennai-born Aravind Srinivas has introduced “Perplexity Computer,” a unified agentic platform that marks its most aggressive expansion beyond the search bar to date. Billed as a “general-purpose digital worker,” the system orchestrates 19 different AI models—including Claude to Gemini—to execute complex, multi-stage workflows such as coding applications, managing research projects, and deploying software.
The launch comes at a precarious moment for the AI industry, as investors begin to demand tangible productivity gains from billions poured into large language models. For Perplexity, it is a bid to evolve from a “knowledge engine” into an “action engine”.
“Perplexity Computer unifies every current capability of AI into a single system," Srinivas, who is also the chief executive officer of the company, wrote in a blog post. “Files, tools, memory, and models, orchestrated together, working for you.”
Beyond the search bar
Since its founding in 2022, Perplexity has carved out a niche by providing concise, cited answers to queries, avoiding “blue links” and ad clutter of traditional search. That formula helped the company reach an annualised revenue run rate of ~$200 million by late 2025, fueled largely by $20-a-month subscriptions.
{{/usCountry}}Since its founding in 2022, Perplexity has carved out a niche by providing concise, cited answers to queries, avoiding “blue links” and ad clutter of traditional search. That formula helped the company reach an annualised revenue run rate of ~$200 million by late 2025, fueled largely by $20-a-month subscriptions.
{{/usCountry}}However, the search market remains a battlefield. Google has fought back with its own AI Overviews, and OpenAI’s SearchGPT has intensified the race for the AI search crown. Perplexity’s response is to move up the value chain. While search provides information, “Computer” aims to provide deliverables.
The system functions by breaking a high-level goal into subtasks. It then assigns those tasks to specialist models: Gemini for deep research, Grok for lightweight operations, and OpenAI’s latest models for long-context recall.
The $200-a-month bet
The new product also serves as a test for Perplexity’s newest revenue tier. Initially, “Computer” is available only to “Perplexity Max” subscribers, a $200-per-month plan launched last year for power users and enterprises.
Crucially, Perplexity is introducing a usage-based credit system for “Computer”, moving away from “all-you-can-eat” subscription model common in the industry. This shift reflects the high compute costs of running nearly 20 models in parallel to complete a single project.
The stakes for the launch are high. Perplexity’s valuation has rocketed from $500 million to $20 billion in just 18 months, supported by backers including Nvidia Corp., Jeff Bezos, and SoftBank Vision Fund 2. To justify that 100X revenue multiple, the company must prove that AI is no longer just a better way to find information, but a replacement for the labour of organising it.
As the industry shifts toward Agentic AI—systems that can navigate the web and use software like humans—Perplexity is hoping its model-agnostic approach will provide a “safe harness” for the next generation of digital work. Whether users are willing to pay a premium for a virtual coworker remains to be seen.