Perplexity Computer adds Plan Mode

// 1h agoPRODUCT UPDATE

Perplexity Computer adds Plan Mode

Perplexity has released "Plan Mode" for its agent-based system, Perplexity Computer, requiring user approval before executing multi-step workflows. This human-in-the-loop design aims to make AI agent actions more transparent, controllable, and reliable.

// ANALYSIS

Plan Mode highlights a crucial UI/UX transition in agentic AI from fully autonomous execution to guided workflows.

–Requiring approval for planned steps mitigates risk in high-stakes workflows like codebase modifications or API interactions.
–Providing users with the ability to edit the plan before execution addresses the unpredictability of modern LLMs.
–This mirrors features like Claude Code's task breakdown and Cursor's multi-file edits, establishing plan-approval as a standard pattern for developer and power-user AI tools.

// TAGS

perplexityagentproduct-updateagentic-workflowshuman-in-the-loopperplexity-computer

DISCOVERED

1h ago

2026-06-12

PUBLISHED

1h ago

2026-06-12

RELEVANCE

8/ 10

AUTHOR

AravSrinivas

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

LAUNCH12m ago

Hellyeah AI launches autonomous growth engine

Hellyeah AI is building AI-native growth and marketing infrastructure that automates campaign preparation, creative generation, and marketing distribution. By integrating agentic workflows that analyze a product, generate assets, and structure ad campaigns, it serves as an autonomous growth engine, helping developers and coding agents overcome the hurdle of product discovery.

NEWS13m ago

Mintlify, Reducto host agent-era fireside chat

Mintlify and Reducto have announced a co-hosted happy hour and fireside chat on June 29th during the AI Engineer World's Fair. The event will feature Hahnbee Lee and Raunak Chowdhuri discussing the design paradigms and practical realities of building developer tools for autonomous AI agents.

INFRA15m ago

Claude Fable 5 prompt caching supports 1-hour TTL

Developer Rhys Sullivan clarified that Anthropic's Claude Fable 5 API natively supports a 1-hour prompt caching Time-To-Live (TTL) duration, debunking assumptions of a strict 5-minute limit. Although CLI environments like Claude Code optimize for 5-minute intervals, the API refreshes the 1-hour cache with each subsequent request.