Perplexity has launched a new AI agent called Computer, which coordinates 19 different AI models to complete complex tasks. The product is currently available to Perplexity Max subscribers for $200 per month. According to Perplexity CEO Aravind Srinivas, Computer “unifies every current capability of AI into a single system” and treats models as interchangeable tools.
The launch of Computer comes at a time when the AI industry is grappling with the question of who will capture the value of AI: the model makers, such as OpenAI, Anthropic, and Google, or the companies that sit above them and turn raw intelligence into reliable products. Perplexity is making a $20 billion bet on the latter. Computer functions as a general-purpose digital worker, accepting high-level objectives from users and delegating subtasks to the best-suited AI model. The platform’s central reasoning engine runs on Anthropic’s Claude Opus 4.6, while Google’s Gemini powers deep research queries and OpenAI’s GPT-5.2 manages long-context recall and expansive web search.
The product’s launch also highlights the risks of local-access models, such as OpenClaw, which can create unnecessary risk and damage data or expose sensitive information. In contrast, Computer runs entirely in the cloud, operating in a safe and secure development sandbox. Perplexity argues that its cloud-based approach is safer and more accessible than local-access models. The company’s revenue grew faster than its user base in 2025, with a 4.7x increase in revenue compared to a 3.7x increase in users.
The launch of Computer has significant implications for the future of AI, with Perplexity betting on the orchestration layer rather than the model layer. The company is positioning itself as an abstraction layer for AI, similar to the companies that built the best abstraction layers above commodity infrastructure in the early days of cloud computing. However, the risk remains that model makers could restrict API access or degrade service to platform competitors. Perplexity is planning to share more details about its search API, ranking embeddings, and infrastructure powering its orchestration stack at a developer event on March 11.
Leave a Reply