Perplexity Launches Upgraded Comet Assistant for Enhanced Web Automation
Perplexity has rolled out a redesigned Comet Assistant that can handle longer, more complex online workflows, automate across multiple tabs, and improve task completion reliability.
By Daniel MercerEdited by Maria Konash
Published:
Updated:
Perplexity upgrades Comet Assistant for advanced browser-based tasks. Photo: Igor Omilaev/Unsplash
Perplexity has introduced a major update to the Comet Assistant, the integrated helper within its Comet browser. Since debuting in July, the Assistant has seen growing adoption for task automation, from travel research to job management and online shopping. The latest release is designed to improve performance on longer, multi-step workflows that span extended web sessions.
The upgraded Assistant features new capabilities for navigating complex online environments, enabling more accurate interaction with dynamic webpages. This approach is intended to help more websites work seamlessly with agent-driven automation while reducing the need for site-specific adjustments.
With enhanced multitasking, Comet Assistant can gather and apply information across several tabs simultaneously. For tasks such as populating spreadsheets with information sourced from multiple websites, users no longer need to guide the Assistant step-by-step. The system can independently access relevant sources and execute workflows in parallel.
Perplexity reports a 23 percent improvement in task success rates compared to the previous version. This increase reflects a stronger ability to follow long instructions, maintain context, and complete sequences requiring precise action ordering. Common tasks highlighted include finding job openings across professional platforms, searching for competitive travel pricing, and building structured datasets like student attendance logs.
Designed for User Choice and Transparency
Perplexity continues to position user control as a core principle of Comet. When a task would benefit from interacting directly with the browser, the Assistant will first request permission and then adhere to that preference throughout the job. The company says this approach ensures transparency while enabling users to decide how much automation they want.
The update aligns with Perplexity’s broader vision to make the internet more efficient, personalized, and less dependent on manual tab-driven workflows. By providing an always-available agent that can execute practical tasks, the company aims to shift browsing from passive search to actionable productivity.
Daniel Mercer covers foundation models, generative AI systems, and applied machine learning deployments across enterprise and consumer platforms. He reports on model launches, performance benchmarks, inference pricing, and the competitive dynamics shaping leading AI labs. His work evaluates compute efficiency, data sourcing strategies, and integration pathways that determine scalability. Daniel takes a systems-level approach, connecting technical documentation and release notes to business outcomes and adoption signals. He frequently analyzes safety mechanisms, model limitations, and reliability claims through testing data and operational evidence. Based in San Francisco, he spends his free time studying urban design and cycling.
OpenAI is leaning toward pushing its IPO to 2027 rather than accept less than a $1 trillion valuation, weeks after filing confidentially for a US listing.
