

Mobile Agent E
Overview :
Mobile-Agent-E is a mobile assistant based on large multimodal models (LMM), aimed at helping users efficiently complete complex multi-step tasks. It achieves self-evolution via a hierarchical multi-agent framework, allowing it to learn and improve from past tasks. The main advantages of this product lie in its powerful reasoning capabilities and competence in handling complex tasks, particularly those involving long durations and interactions among multiple applications. It is suitable for users who require efficient completion of complex mobile tasks, such as business professionals and researchers. Currently, it is in the research phase, and specific pricing has not been established.
Target Users :
This product is designed for users who need to efficiently complete complex tasks on mobile devices, such as business professionals, researchers, and students. It helps users reduce manual operations and enhance task completion efficiency, especially in scenarios requiring cross-application interaction and long-term planning.
Use Cases
Complete a restaurant recommendation task on a mobile device, including searching, filtering, and making reservations.
Shop online by comparing products across multiple applications and completing purchases.
Plan a travel itinerary involving queries and reservations across several applications.
Features
Hierarchical multi-agent framework including managers, perceivers, operators, action reflectors, and recorders.
Self-evolution module enhances performance through cues and shortcuts in long-term memory.
Handles complex tasks involving long durations and interactions among multiple applications.
Introduces a new evaluation metric—Satisfaction Score (SS)—closer to human preferences.
Provides detailed error feedback and an automatic adjustment mechanism to improve task success rates.
Supports various large multimodal models as backends, such as GPT-4o and Gemini.
Validates performance through Mobile-Eval-E benchmarking, significantly outperforming existing methods.
Offers open-source code and datasets for research and development purposes.
How to Use
1. Visit the project homepage to understand the product features and architecture.
2. Download the open-source code and follow the guidelines for local deployment.
3. Prepare task inputs, such as task descriptions and initial application interfaces.
4. Run Mobile-Agent-E and observe the high-level plans and execution actions it generates.
5. Adjust model parameters or long-term memory content as necessary to optimize performance.
6. Use the Mobile-Eval-E benchmark to assess model performance.
7. Expand or modify model functionalities based on actual needs to cater to specific tasks.
Featured AI Tools
English Picks

Popai
PopAi is a product providing AI assistant services, integrated with GPT-3.5 technology. It offers powerful chat, document creation, and creative generation capabilities. Users can interact with AI by uploading files or links, or leverage AI to assist with tasks like educational writing, professional writing, presentation creation, and programming problem-solving. PopAi aims to enhance user productivity and creativity, offering a superior AI assistant experience.
Personal Assistance
1.7M

Named By AI
AI naming is an intelligent name tool that uses artificial intelligence to help you find unique and meaningful names for your baby. You can choose the baby's gender, name origin, name theme, and popularity, and AI naming will generate a series of excellent names based on your preferences and tastes.
Personal Assistance
1.1M