Perplexity Just Built an AI That Does Everything
TL;DR
Perplexity Computer is pitched as hosted OpenClaw for normal humans: Berman says his OpenClaw install was powerful but expensive and high-maintenance, while Perplexity Computer delivers similar agent behavior out of the box.
Connectors are the standout feature: Instead of manually wiring tools like Telegram, Asana, Gmail, and Calendar, Perplexity offers hundreds of pre-built integrations including Drive, Notion, GitHub, Linear, Dropbox, and Gmail with simple authentication.
It runs real agent workflows with code execution and sub-agents: He shows tasks running in parallel, writing terminal commands, delegating work internally, and scheduling future actions like a UFC briefing that wakes up Saturday at 1:00 p.m. PST for a final sweep.
Skills make custom automations portable: Berman quickly recreated his OpenClaw food journal, where an uploaded burger photo became a logged meal entry with ingredients and timestamp after loading a saved skill.
The pricing is usage-based and visible per task: He cites 15 credits for checking his next calendar meeting, 388 credits for a Gemini 4 benchmark chart with image output, and 1,500 credits to install his knowledge base kit.
The killer use case is a persistent searchable knowledge base: In about five minutes, Perplexity built a web app that ingests links, turns them into embeddings, stores them in a database, and lets him and his team search past articles, videos, and tweets with natural language.
The Breakdown
Matthew Berman says Perplexity Computer gives most people the power of his heavily customized OpenClaw setup without the painful setup, security tradeoffs, or constant maintenance. He shows it rebuilding his real workflows, from a UFC fight-night brief to a food journal and a persistent team knowledge base, while exposing the exact credit costs along the way.
Was This Useful?
Share
Keep Reading
Make Alcreon Yours
Tune your feedFive quick questions, and the feed ranks what matters to you first.Or just get notified
The weekly Echo. Signal worth keeping in your inbox.
Every new piece, announced on X.
Read Next
See all
Playbook
Cheap Models, Hard Tasks
Most agent workflows route every step to the frontier model by default. The bill scales with how chatty the agent gets, even when most steps don't need that brain.
Playbook
Tasteful Skills
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.
Playbook
The Art of Tasteful Prompting
Learn how tasteful prompting helps you move beyond generic AI output by shaping context, style, and judgment from the start.