Quick Start

1. Overview

AgentWeave is a powerful AI collaboration platform designed to help users quickly create, manage, and deploy AI agents. Through simple operations and rich features, users can easily build intelligent agents suitable for various scenarios.

2. Getting Started

2.1 Account

2.1.1 Register and Log in to Your Account

1. Open your browser and visit the AgentWeave official website.
2. Click the "Register" button on the page and fill in the necessary registration information (such as email, password, etc.).
3. After completing the registration, log in to the platform using your registered email and password. Alternatively, you can choose to log in using Google or Apple authentication.

2.1.2 Recover Password

1. Open your browser and visit the AgentWeave official website.
2. Click the login button in the upper right corner to enter the login page. At the bottom of the page, you'll find a "Forgot Password" option.
3. Click to enter the forgot password page and follow the on-screen instructions to enter your email address. Click to send a verification code.
4. After successfully sending, check your email for the verification code and enter it.
5. Enter your new password and click submit. You have now completed the password recovery process.

2.1.3 Account Dropdown Menu

After successfully logging in, clicking on your avatar in the top right corner will reveal the following list:

1. Edit Profile

In the personal profile editing interface, you can update the following basic information: avatar, nickname and signature.

2. My Files

Clicking on "My Files" allows you to view the files you've uploaded to the system, including information such as file size and upload time.

3. My Collections

You can add agents to your favorites by clicking the "Add to Favorites" button on the agent details page. This includes the default favorites folder, personally created folders, and shared folders.

4. My Feedback

Feedback submitted through the feedback channel and the results will be displayed on this page.

5. My Activity

At the bottom of the page, you can search by date and agent name to view the usage time, messages, tokens, and speed of relevant agents.

6. My Credits

Displays the credits balance and total cash balance, as well as details of credit and cash credit records.

7. Upgrade Plan

Clicking this will redirect you to the pricing page, where you can purchase personal or business plans.

8. Agent Transfer Records

Click to view records of agents transferred by you or received by you.

9. Settings

If you've purchased a personal plan, you can cancel it here.

You can also modify your password.

2.1.4 Change Password

1. Open your browser, visit the AgentWeave official website, and log in successfully.
2. Click on your avatar in the top right corner and select "Settings" from the dropdown menu.
3. On the Settings page, there will be a "Change Password" button. Click it to enter the password change page.
4. Follow the prompts to enter your old password and a new password that meets the password requirements. Click "Change" to reset your password.

3. User Guide

3.1 Introduction to AI Agent Templates

3.1.1 Agent Templates

1. Conversational & Interaction

Purpose: Designed for natural, open-ended dialogues. Build agents for chat, customer support, quick information queries, or light tasks. Users can choose different large language models and seamlessly integrate with built-in tools to enhance workflows.

LLM-Powered Chat Builder

1. Personalized AI

Purpose: Suitable for building personalized intelligent assistants or dedicated agents. Can be customized around users' personal needs, interests, or work tasks for efficiency improvement, habit formation, learning guidance, or life management.

AI Personal

1. Knowledge & Retrieval

Purpose: Suitable for scenarios that require processing and calling large amounts of documents, databases, or repositories. Through Retrieval-Augmented Generation (RAG), users' agents can provide precise answers based on existing knowledge, ideal for applications such as FAQs, knowledge base Q&A, report generation, and document assistance.

Google Vertex AI RAG

AWS Bedrock RAG

Bedrock RAG - S3 Vectors

OpenAI RAG

Microsoft RAG

Users can choose based on their needs to create their own agents. The agent creation process is described below.

3.1.2 AI Building

In the description box, specifically describe the agent you want to build. The AI will analyze your description and generate relevant content for the agent, including the agent template, name, description, etc. Except for the template, the rest of the content can be modified on the management page.

3.1.3 Agent Team

Combined agents are systems based on multi-agent collaboration, typically used for automated collaboration on complex tasks. Through collaboration between agents, deep information processing and improved accuracy can be achieved. For example, in knowledge Q&A and document parsing scenarios, the division of labor and collaboration between agents can enhance the depth and accuracy of information processing.

AutoGen Agent

AWS Step Function

A2A Team

3.1.4 Agent Space

You can establish an agent space using existing templates or customize your own. In the agent space, you can add multiple different agents for discussions on specific topics. Users can obtain diverse insights from multiple AIs in a single conversation interface.

3.2 Creating an AI Agent

3.2.1 Selecting a Template

1. After logging in, enter the AgentWeave workspace.
2. On the workspace page, find the "⊕" icon and click to enter the template selection page.
3. AgentWeave offers various preset templates covering common application scenarios. Choose a template that suits your needs, such as the "OpenAI RAG" template.
4. Select the template to enter the template details page.

3.2.2 Basic Information

1. After clicking the "Create Agent" button, enter the agent configuration page.
2. Fill in the agent's basic information:
   1. Agent Name: Enter an easily identifiable name, e.g., "My Customer Service Bot".
   2. Who Can See: Choose the visibility settings for the agent.
   3. Share: By clicking the "Select" button, you can specify which users or organizations can access this agent. This helps control who can interact with the agent.
   4. Owner: Select the owner of the agent, which can be an individual or an organization. For team collaboration projects, it's recommended to choose an organization as the owner so team members can jointly manage and use the agent.
   5. Introduction: Briefly describe the agent's functions and purpose.

After completing the above information, click the "Next" button in the upper right corner of the page to continue with further agent configuration. This will ensure your agent is properly set up and ready for subsequent customization and deployment.

3.2.3 Configuring the Agent

On the agent configuration page, you can make relevant settings according to business needs, including whether to charge users, the number of credits required per query, opening remarks, prompts, etc. Specific configurations may vary under different templates; refer to the agent section introduction for detailed content.

After completing the above configuration, click the "Create" button in the upper right corner of the page to save the settings and complete the agent creation.

3.2.4 Writing Prompts

Configuring a more suitable agent requires writing prompts, i.e., the agent's "Persona & Prompt". The agent's persona and prompt define its basic role setting and will continuously influence its response style in all conversations. It is recommended to clearly define the model's role in the prompt, design the language style of responses, and limit the model's response range to make the chat more in line with user expectations. (After the agent is published, you can enter the management interface to modify and write prompts.)

1. Importance of Prompts
   1. Prompts are key to guiding the agent's behavior and output. A well-written prompt ensures that the agent maintains the expected role, style, and content range when interacting with users. This not only enhances user experience but also effectively avoids irrelevant or inaccurate responses from the agent.
2. Components of Prompts
   1. Persona Definition: Clearly define the agent's identity and responsibilities. For example, the agent can be a "Q&A Expert", "Creative Writing Assistant", or "Data Analyst".
   2. Task Instructions: Detailed explanation of the specific tasks the agent needs to complete. This includes how to process input, how to think, and how to generate output.
   3. Output Format Requirements: Specify the structure and format of the agent's responses. For example, whether credits need to be listed, whether summaries are required, whether specific markers are needed, etc.
   4. Constraints and Limitations: Clearly state the rules the agent needs to follow when responding, such as only using information from search results, needing to mark conflicting information, etc.
3. Suggestions and Best Practices for Writing Prompts
   1. Clarity and Specificity: Avoid vague expressions and use specific, clear language. For example, instead of saying "answer questions", say "use information from search results to answer user questions".
   2. Role First: Clearly define the agent's role at the beginning of the prompt, which helps the agent quickly get into character.
   3. Step-by-Step Guidance: Break down complex tasks into simple steps and arrange them in logical order, guiding the agent to complete tasks step by step.
   4. Limit Response Range: Further restrict the agent's response range by explicitly prohibiting certain types of content or behavior to ensure it meets expectations.
   5. Iteration and Optimization: Writing prompts is an iterative process. In actual use, you may need to continuously adjust and optimize prompts based on the agent's performance to achieve the best results.

By clearly defining roles and rules, combined with testing and iteration, you can make AgentWeave's intelligent agent's "Persona & Prompt" precisely serve the scenario, making conversations both "have identity" and "meet needs", continuously improving the user interaction experience.

3.3 Publishing Agents

3.2.1 Publication Process

1. Submit for Review: When you click the "Publish" button, your agent will be submitted to the platform for review. This is an automated process designed to ensure all agents meet the platform's standards and policies.
2. Await Review: The review process may take some time, depending on the platform's workload and review queue. During this period, your agent will not be publicly accessible.
3. Review Results: Once the review is complete, you will receive a notification. If the review passes, your agent will be activated and available for users to access. If the review doesn't pass, you will receive feedback notification and need to make corresponding modifications based on the feedback.

3.2.2 Post-Publication Steps

1. Try It Out: After the agent is successfully created, you can click the "Try It" button to test your agent and ensure it works as expected.
2. Manage: Click the "Manage" button to enter the agent's management interface, where you can perform further configuration and monitoring.

3.3 Agent Management

3.3.1 Managing Agents

1. Access the "My Agents" page in the agent management interface:
   1. After logging into your AgentWeave account, navigate to the "My Agents" page. This lists all the agents you've created.
   2. Select an agent: Find the agent you need to manage from the list and click to enter its details page.
   3. Click the "Manage" option: On the agent details page, find and click the "Manage" option. This will guide you to the agent's management interface where you can perform various management operations.
2. In the management interface, you can access different management functions through the left navigation bar:
   1. Dashboard: View the agent's performance metrics and user interactions.
   2. Files: Manage files and knowledge base content related to the agent.
   3. Usage History: View the agent's operation history and user interaction records.
   4. Settings: Adjust the agent's configuration, such as permissions, visibility, and feature settings.
   5. Conversations (only visible if enabled during creation): Search conversations by date, user, keywords, etc.
   6. Ownership: The agent owner can transfer ownership of the agent to an individual or organization.
   7. API Key: Manage the agent's API access keys for integration and automation.

Through these steps, you can effectively manage and edit your agents, ensuring they meet your business needs and provide the best user experience.

4. Pricing Plans

4.1 Upgrading Plans

Users can upgrade their AgentWeave plan at any time based on business or usage needs to get more quotas, stronger features, and collaboration capabilities. Here are the steps to upgrade your plan:

1. Log in to the AgentWeave platform
   1. Use your account to log in to the AgentWeave platform.
2. Enter the "Pricing" page
   1. Click "Pricing" in the top navigation bar to enter the plan selection interface.
3. Choose a suitable plan
   1. New users who register for the first time have a 14-day free trial of the personal plan.
   2. Select a plan type based on your needs: Personal Plan ($10/month): Suitable for individual high-frequency use. Pro Plan ($20/month): Suitable for small teams and professional users. Team Plan ($220/month): Suitable for large teams or organizations that need extensive collaboration, high usage, and dedicated agent resources. Team Plus ($550/month): Suitable for large teams or organizations that need extensive collaboration, high usage, and dedicated agent resources. Enterprise Plan: Suitable for large organizations, supporting private deployment and customized services.
4. Click the "Subscribe" or "Contact Us" button
   1. Click the corresponding button based on your chosen plan to complete payment or submit an enterprise inquiry.
5. The system automatically takes effect after successful upgrade
   1. The upgrade takes effect immediately, and you can view your current plan and next billing date in "Settings".

4.2 Cancelling Plans

When users no longer need the current plan service, they can perform self-service cancellation in the system. After cancellation, there will be no automatic charging for the next cycle, but you can still use the activated features normally within the current cycle. Here are the steps to cancel a plan:

1. Log in to the AgentWeave platform.
2. Click on your avatar in the top right corner to enter the "Settings" page.
3. Find your current subscription information (such as Personal Plan, Pro Plan, etc.) on the "Settings" page.
4. Click the "Cancel Subscription" button on the right.
5. After confirming the cancellation, the system will stop automatic charging for the next cycle.
6. The page will display the next billing date (e.g., 2025/07/04 14:23) for reference.