Perplexity Just Launched Their Own OpenClaw

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

Summary

Perplexity has introduced "Perplexity Computer," a multi-model agentic system designed to autonomously decompose prompts into tasks, orchestrate specialized sub-agents, and execute workflows end-to-end. This system utilizes 19 AI models, with Claude Opus 4.6 serving as the core reasoning engine, and features hundreds of connectors, persistent memory, and web access, all within isolated sandboxed environments. Tasks can run for extended periods, checking in only when human intervention is genuinely required. Concurrently, Cursor has launched an updated version of its cloud agents, which now operate within dedicated virtual machines, enabling parallel execution and self-validating output through interaction with the software they build. These agents are accessible across multiple platforms and can be remotely controlled, with Cursor internally reporting over 30% of merged PRs are agent-generated. Additionally, Alibaba's Qwen team released the open-weight Qwen 3.5 Medium Series, featuring a 35B model that activates only 3B parameters per token yet outperforms a previous 235B flagship model.

Key takeaway

For CTOs and VPs of Engineering evaluating AI agent adoption, Perplexity Computer and Cursor's enhanced cloud agents demonstrate a shift towards more autonomous, robust, and secure agentic workflows. You should consider integrating these advanced agent systems to offload complex development tasks, automate code generation, and improve testing efficiency, leveraging their sandboxed execution and multi-model orchestration capabilities to accelerate project delivery and reduce manual overhead.

Key insights

Autonomous AI agents are evolving to handle complex, long-horizon tasks within isolated, multi-model environments.

Principles

Method

Perplexity Computer decomposes prompts, routes subtasks to 19 models via Claude Opus 4.6, and executes in sandboxed environments with persistent memory. Cursor agents run in isolated VMs, build/test software, and deliver validated artifacts.

In practice

Topics

Code references

Best for: NLP Engineer, CTO, VP of Engineering/Data, Machine Learning Engineer, AI Engineer, Software Engineer

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by unwind ai.