Reviewed by Jonathan West · Updated Sep 28, 2026

How to Use Manus AI for Complex Workflows

A walkthrough of agent execution modes and automation workflows across free and paid plans.

Reviewed by Jonathan West · Updated Sep 28, 2026

To use Manus AI, sign up, then give Agent Mode one concrete goal and name the file you want back.

At Layer3Labs, we build AI workflows for small and mid-sized businesses, so the steps below are how we would set up a first Manus task, not Manus's own instructions.

Manus does the work in a cloud virtual computer, and each Agent Mode task spends credits from your daily or monthly allowance.


How to Use Manus AI: Your First Task

You use Manus AI by asking quick questions in Chat Mode, which costs no credits, and giving multi-step jobs to Agent Mode. Scheduled tasks rerun a job on a timetable, and Playbooks give you a template to start from.

Write a precise first prompt so Manus can plan the job. State the target deliverable, provide reference documents or data, and define the file format you expect at completion.

  • Account setup: Create an account on the Manus registration page to receive 1,000 starter credits.
  • Mode selection: Select Chat Mode for conversational queries without spending credits, or choose Agent Mode to create slides, code, and web applications. You can switch an active conversation from Chat Mode to Agent Mode, but Manus does not allow switching back.
  • Prompt drafting: State a single concrete goal alongside your exact deliverable structure, such as a CSV spreadsheet, a formatted PDF document, or a functional web page.
  • Plan observation: Watch Manus plan the steps and carry them out in its cloud virtual computer.
  • Output review: Inspect the generated files in the workspace viewer before downloading or deploying.
  • Credit reconciliation: Check your remaining credit balance in your account settings, as Agent Mode deducts credits for each action upon starting the following action.

Allowance Limits and How to Use Manus AI for Free

The Manus Free plan provides 1,000 one-time starter credits upon registration alongside a daily refresh of 300 credits. These daily credits grant access to basic autonomous runs without paying a monthly subscription fee.

Daily refresh credits arrive upon login and reset every day at 00:00 Coordinated Universal Time (UTC). Daily refresh credits do not roll over. Free accounts can use at most 1,500 daily refresh credits a month, and refreshes stop until the next cycle once they reach that cap.

Daily credits can only be used in Lite mode. The help center lists Manus 1.6 Lite for free agent runs. The tooltip on the Manus pricing page says Manus 2.0 Lite, and Manus 2.0 launched on September 28, 2026. Manus may rename the modes after the 2.0 launch. Free accounts are limited to 1 concurrent task and 2 scheduled tasks.

Chat Mode consumes zero credits on both Free and paid accounts. Use Chat Mode for web lookups and document reading to preserve your daily credit allowance for code execution and multi-step research.

Wide Research Execution and Subtask Scaling

Manus Wide Research is a paid-only feature that splits a large research job into parallel subtasks and can run 20 of them at once. Manus caps each subtask at 50 credits.

Wide Research supports only automatic triggering, according to the Manus Wide Research documentation. You cannot switch Wide Research on by hand.

The architecture relies on a coordinator model. The main agent splits your goal into research questions and gives each one to a sub-agent with a fresh context. It then combines the findings into one report.


Manus Scheduled Tasks and the Automations Engine

Manus Automations expand scheduled tasks by allowing runs to trigger from third-party app events as well as recurring clock intervals. With the launch of Manus 2.0, older scheduled tasks continue to operate inside this unified Automations system.

Trigger sources include Gmail, Outlook Mail, Mail Manus, Notion, Google Calendar, Outlook Calendar, GitHub, Shopify, Google Workspace, advertising services, RSS feeds, and incoming Webhook payloads. GitHub, Shopify, Webhook, and Mail Manus events trigger runs in real time, while other apps are checked at intervals, so their triggers may not start instantly.

The Manus Automations guide lists three automation types.

  • Schedule: starts a task at a set time.
  • Triggered task: starts when something happens in a connected app.
  • Advanced automation (beta): built from a plain-English request.
  • Free tier limits: Free users can configure up to 2 scheduled tasks and run 1 concurrent task.
  • Paid tier limits: Paid subscriptions support up to 20 scheduled tasks and 20 concurrent tasks.
  • Continuous runs: Since Scheduled Tasks 2.0, each run continues inside the same task, so you review every run in one place.
  • Confirmation bypass: You can enable a skip-confirmations setting to let recurring automations run without manual prompts.

Manus Playbook Templates Compared to Scheduled Tasks

A Manus Playbook is a pre-built workflow template, whereas an automation or scheduled task is an execution rule that defines when and how a prompt runs.

The Manus Business Playbook directory offers a public gallery of templates for business teams. Examples include market research builders, business plan writers, and investor pitch deck makers. Each Playbook gives you a template to start a task from instead of a blank prompt.

Our suggestion: start a task from a Playbook, then set up a Schedule with the same prompt if you want it to rerun. Manus maintains user privacy by ensuring that private projects, templates, and customized workflows are never displayed publicly without explicit user consent.


Alternative Input Channels from Email to Slack

You can send tasks to Manus through email, Slack channels, browser extensions, and native desktop software. These alternative interfaces allow team members to assign work to Manus without opening the primary web browser dashboard.

Mail Manus assigns an @manus.bot email address to your account. You send requests directly from up to 10 pre-approved sender addresses. Workflow Emails apply your custom instructions to whatever you send them. Free users can have up to 3 Workflow Emails and make 10 creates or deletes a day. Paid users can have up to 30, with 60 changes a day. Custom email addresses can be changed once every 7 days.

Teams using Slack can install the integration from account settings and invoke @manus in project channels or direct messages. The Manus Slack page says you need a paid Slack plan for its AI features, so on a free Slack workspace use Mail Manus instead. For local files and terminal commands, install the Manus desktop app on an Apple Silicon Mac or a Windows PC. To let Manus work in your own logged-in browser, add the Browser Operator extension to Chrome or Edge.


Credit Spend Safeguards and Refund Policies

Manus consumes account credits dynamically based on Large Language Model (LLM) tokens, Virtual Machine (VM) runtime, and external Application Programming Interface (API) calls. Complexity and processing duration determine the total credits deducted for any given job.

Manus deducts credits incrementally per action. The cost of an action is calculated and billed at the start of the following action. This billing sequence means your available balance can momentarily drop below zero during an active run. Negative balances clear automatically during your next credit refresh, but Manus blocks new tasks from starting while your credit balance remains at or below zero.

Manus cannot provide accurate credit estimates before running a task. According to the Manus credit documentation, any credit prediction Manus generates inside a chat should be treated as an AI hallucination. However, the Manus refund policy provides a full credit refund for tasks that fail due to platform technical issues, tool misuse, or incomplete execution. To learn what a job costs, run a small version first and compare your balance before and after.

Credit refunds do not cover vague instructions, subjective dissatisfaction, user changes of mind, or third-party website failures. Be specific in your prompt so Manus does not spend credits on steps you did not want.

Workflow Selection and How to Use Manus AI Effectively

Manus AI is built for multi-step jobs like research briefs, slide decks, spreadsheets and websites. Those jobs need a browser, code and files, which Manus's virtual computer provides.

Manus is not suitable for work you cannot independently review or verify. Do not assign Manus unattended database modifications, live production code deployments, or critical legal compliance filings where an unnoticed hallucination carries operational liability.

Our operational guidance would change if Manus introduces hard user-defined credit spending caps per task, or allows unspent daily refresh credits to roll over across billing cycles. To use Manus AI without surprises, run one small Agent Mode task that names its output file, then read how Manus bills credits before you schedule it.

Frequently Asked Questions

  • You use Manus AI for free by registering an account to receive 1,000 starter credits, then relying on the 300 daily refresh credits granted upon login. Free accounts can execute tasks in Chat Mode without consuming credits, or run autonomous tasks in Lite mode. Daily credits reset at 00:00 UTC, and Free accounts can use at most 1,500 daily refresh credits a month.
  • No, you do not have to pay to use Manus AI. The Free plan provides access to Chat Mode and Manus 1.6 Lite in Agent Mode with 300 daily refresh credits, 1 concurrent task, and 2 scheduled tasks. Paid plans starting at $20 monthly are only required if you need Wide Research, higher credit allowances, 20 concurrent tasks, or access to flagship agent modes.
  • Manus AI should be used for complex, multi-step tasks that require web browsing, data extraction, file generation, and code execution. Typical jobs include multi-source market research, slide deck preparation, spreadsheet compilation, and web application prototyping. It should not be used for tasks where you cannot independently verify the resulting code or data.
  • A Manus Playbook is a structural template that outlines the prompts and workflow steps for a specific business task, while a scheduled task is a time-based execution trigger. You use a Playbook to define what Manus should build, and you use a scheduled task or automation to determine when Manus runs it.
  • Manus Wide Research is an automated, paid-only feature that splits extensive research questions into parallel sub-agents, 20 at a time. Each sub-agent investigates a specific angle with a fresh context window, capped at 50 credits per subtask, before a lead agent synthesizes their findings into a single structured report.
  • You schedule a task in Manus by configuring an entry in the Automations dashboard and choosing a recurring time schedule or external trigger. Free accounts can set up to 2 active scheduled tasks, while paid subscribers can maintain up to 20 scheduled tasks simultaneously.

Audit Your AI Automation Workflows

We map which of your workflows fit an agent and where a person should check the output.

Book a Consultation