I didn't think this was possible...
TL;DR
Perplexity Computer is pitched as hosted OpenClaw: Berman says it offers the same core agent behavior, code execution, web access, and multi-step workflows as OpenClaw, but without the maintenance burden that had him spending more time babysitting the system than getting value from it.
Prebuilt connectors are the standout feature: Instead of manually wiring Telegram, Asana, Gmail, Calendar, Notion, GitHub, Dropbox, and other services, Perplexity ships with hundreds of connectors that authenticate like normal apps and avoid the DIY API-key setup he had to do in OpenClaw.
Threading and parallel task execution make it practical: Tasks are organized as separate threads by default, and Perplexity can run multiple threads and sub-agents in parallel, which Berman says keeps context cleaner and mirrors how serious agent workflows should work.
He recreated real workflows in minutes: Examples include checking his next calendar event, building a UFC fight-night brief for Saturday at 1 p.m. PST, recreating his stomach-tracking food journal from a burger image, and generating a benchmark chart for Gemma 4 versus Qwen 3.6, MiniAX, and Kimi K2.5.
The biggest proof point was rebuilding his team knowledge base: In about five minutes, Perplexity planned and built a persistent web app that ingests links, converts them into embeddings, stores them in a searchable database, and can later be exposed through channels like Telegram or WhatsApp.
Pricing is usage-based on top of a Perplexity plan: He shows example costs of 15 credits for a simple calendar lookup, 388 credits for the Gemma 4 benchmark chart with image output, and 1,500 credits to install his knowledge-base kit, giving a rough sense of which tasks are cheap versus expensive.
The Breakdown
Perplexity Computer rebuilt Matthew Berman's most valuable OpenClaw workflows, including a food journal, earnings reports, and a persistent knowledge base, without the setup pain, token burn, and connector headaches that made his self-hosted stack hard to justify. His verdict is blunt: if you want an OpenClaw-style agent that "just works" for most people, this is probably the better option.
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
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.

Playbook
The Codex /goal Playbook
OpenAI shipped /goal for the Codex CLI. It turns a prompt into a persisted, self-continuing contract.