← Back to home

AI Agent workflow building

Describe the task and let the Agent plan, write and revise the current project.

08

How to use it

State the target application, inputs and acceptance result. The Agent reads project structure, develops a requirement and execution plan, then writes steps into the existing project. Checkpoints and write evidence support recovery and replanning, while the workflow remains editable and testable.

Getting started →

Mouse and keyboard recording

Capture continuous input and tune counts, intervals, hold durations and delays for every action.

Explore the feature →

Multi-template matching and clicks

Configure multiple visual candidates independently, locate a target and click the detected position.

Explore the feature →

Route networks and automatic path planning

Build and save a route network, then plan movement from the current position through required waypoints.

Explore the feature →

AI deep-learning detection

Run a trained model locally and output object regions, classes, confidence and screen coordinates without consuming large-model tokens.

Explore the feature →

MMaxLab training platform

Annotate, augment, train, run inference and validate locally without consuming large-model tokens.

Explore the feature →

Project library and backups

Create, search, favorite, import, export and switch projects. Share exported automation projects for other users to reuse.

Explore the feature →
NEXT STEP

Start with one real task

Install MacroMax, record a short sequence, then add vision and logic as the task grows.

Download for Windows (2.1 GB) ↗