Manus
Manus is an autonomous AI agent: instead of just replying in a chat, it plans a task, takes its own actions to complete it, and hands back a finished result. This guide covers the basics, what it's good at, current plan options, and a few practical ways to keep it on track.
What is Manus?
Manus is a general-purpose AI agent. Give it a goal in plain English and, instead of just answering, it plans the steps needed, browses the web, writes and runs code, reads and edits files, and works through the task largely on its own before returning a finished output like a report, a spreadsheet, a slide deck, or a small website.
That makes it different from a chat assistant you go back and forth with on every step. You describe the destination, and Manus figures out the route, checking back in mainly when it needs a decision from you or when the task is done.
Getting started
- Go to manus.im and sign up.
- Use Chat mode for quick questions, or Agent mode when you want it to actually go do something.
- Describe the end result you want, not just the first step.
- Watch the step-by-step trace as it works, and reply if it asks you a question.
A downloadable slide deck comparing the three competitors.
What Manus can do
These are the outputs it's most often used for. The examples show the sort of request each one handles well.
Research & reports
Hand it a topic and it searches, reads multiple sources, and writes up a structured summary.
Slides & spreadsheets
Turn a request into a finished slide deck or a spreadsheet with real data, not just an outline.
Websites & small apps
Describe a simple site or tool and Manus can build and host a working version.
Data analysis
Upload a file and it can clean, analyze, and chart the data, writing and running its own code.
Pricing tiers
Manus bills usage as credits: every action the agent takes (planning, browsing, running code) spends credits, and complex tasks can use up a lot of them, so costs are less predictable than a flat monthly fee. Check the official pricing page before you rely on it for anything big. The plans mainly differ in how much work they let you run:
- · 300 credits refreshed daily
- · 1 task running at a time
- · Good for trying it out
- · Ladders up to $200/mo for more credits
- · Faster models, more concurrent tasks
- · Credits reset monthly, don't roll over
- · Shared credit pool for the team
- · Admin controls and usage analytics
- · 2-seat minimum
Because credit use scales with how much work a task actually takes, it's worth starting small and checking your credit balance before handing over something large or open-ended.
Prompting tips
- Describe the finished output you want (a deck, a report, a working page), not just the topic.
- Give it any constraints up front: length, format, sources to prefer or avoid.
- For a big task, ask it to check in with a plan before it starts, rather than running the whole thing unsupervised.
- Break a very large task into smaller ones if you want to keep an eye on credit usage.
- If a step looks off partway through, say so immediately instead of waiting for the final result.
More usage details live in Manus's help center.
For longer, multi-step jobs
Manus sits alongside the app builders covered in this AI Tools Course, such as Lovable, but is broader: instead of only building apps, it takes on open-ended research, reporting, and data tasks the same way. After this, the course moves into image tools, then video and audio tools.
What's Next
Next up: moving from AI agents into image generation tools.