Perplexity's 19-model AI ‘Computer’

· Source: The Rundown AI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, Software Development & Engineering · Depth: Intermediate, medium

Summary

Perplexity has launched "Perplexity Computer," a new multi-model AI agent system that orchestrates 19 distinct AI models to autonomously handle complex tasks. This platform allows users to define an outcome, prompting the system to deploy sub-agents for browsing, coding, and app connectivity within a sandboxed environment. Perplexity Computer is designed for long-running operations, potentially active for months, and offers consumption-based pricing with a 10,000-credit monthly bank for its Max tier. Users can also manually select specific models for individual tasks, emphasizing flexibility as a core product feature. This release positions Perplexity as a key player in the agentic AI space, contrasting with single-model approaches from competitors like Anthropic.

Key takeaway

For CTOs and VPs of Engineering evaluating AI agent solutions, Perplexity Computer offers a compelling multi-model, sandboxed approach for long-duration tasks. Your teams can leverage its flexibility to orchestrate diverse AI capabilities, potentially reducing operational overhead and accelerating complex project completion. Consider piloting Perplexity Computer for tasks requiring extensive browsing, coding, or app integration to assess its efficiency and cost-effectiveness compared to single-model agents.

Key insights

Perplexity Computer orchestrates 19 AI models for autonomous, long-running, sandboxed agentic task execution.

Principles

Method

Users define an outcome, then the system spins up sub-agents that browse, code, and connect to apps, orchestrating 19 rival models across tasks within a sandboxed environment for extended periods.

In practice

Topics

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Engineer, AI Product Manager, Marketing Professional

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Rundown AI.