Brightsy AI is an advanced AI Agent Tooling platform designed to streamline your enterprise data flows and create intelligent, autonomous systems. This guide will help you understand the core components and how to get started with Brightsy AI.
Key Concepts
Prompting
Effective prompting is the foundation for success with Brightsy AI. Understanding how to craft clear, specific instructions for AI agents enables optimal performance and results. We recommend the Prompt Engineering Guide as an excellent resource for mastering prompt engineering techniques, including:
- Zero-shot and Few-shot Prompting: Instructing AI with or without examples
- Chain-of-Thought: Guiding AI through complex reasoning steps
- Retrieval Augmented Generation: Enhancing responses with external knowledge
- Advanced Techniques: Leveraging the latest prompting methodologies
For practical examples specific to Brightsy AI tools, see our Prompting Guide with detailed examples and best practices.
Agents
Agents are configurable AI assistants that can be customized with specific instructions, knowledge bases, and tool access. They hold credentials to registered Applications and utilize the associated Tools to perform operations. Agents can be connected to each other, creating a network of specialized capabilities for complex task delegation. Key features include:
- Expert Agent Delegation: Core platform design enabling specialized AI for different tasks
- Multi-Agent Interactions: Agents can collaborate on complex workflows
- Autonomous Operation: Independent decision-making within defined parameters
- Reasoning Capabilities: Transform raw data into actionable insights
- Memory Management: Store and retrieve contextual information
Scenarios
Scenarios are predefined interactions that can be triggered manually, on a schedule, or via webhooks. The scenario combines instructions with incoming data and executes against the assigned Agents. Key features include:
- Automated Workflows: Run predefined processes based on triggers
- Event Response: Process webhooks from third-party services (like GitHub)
- Scheduling: Cron-based scheduling for regular executions
- Webhooks: Receive and process data from external systems
Applications
Applications are integrations with external APIs and services registered using OpenAPI 3.0+ specifications. Each operation defined in the API is translated into a Tool that is registered with the AI model. Key features include:
- Authentication Management: Automatic handling of OAuth2, API keys, and tokens
- Extended Capabilities: Provide agents with access to external services
- API Request Handling: Simplified interfacing with external services
CMS
Define structured data schemas with validation rules and UI configuration. They provide a foundation for data management across the platform. Key features include:
- Schema Definition: Create custom data structures with validation rules
- UI Configuration: Customize forms for data entry and viewing
- Natural Language Search: Semantic search on enabled record types
- Advanced Querying: Complex filtering, sorting, and aggregation
- Vector Database Support: For Retrieval-Augmented Generation (RAG) applications
File Management
The file system provides capabilities for storing, organizing, and accessing documents and media. Key features include:
- Document Organization: Hierarchical folder structure
- Search Capabilities: Find files quickly across the system
- Secure Sharing: Reference files within messages and forms
- Access Control: Permission-based file access
Tools
Tools are operations that enable the AI model to perform specific tasks. They include:
- Data Management Tools: CRUD operations for record types
- File Tools: Upload, organize, access, and share files
- UI Tools: Present interactive components (forms, buttons, charts)
- Integration Tools: Connect with external APIs and services
- Memory Tools: Store and retrieve contextual information
- Web Tools: Make external API requests and process responses
Getting Started
1. Register Applications
Begin by registering Applications using OpenAPI 3.0+ specifications. This automatically generates Tools based on the API operations, giving your Agents access to external services.
2. Create Record Types
Define structured data schemas for your business objects with custom validation rules and UI configurations. These provide the foundation for organized data management.
3. Set Up File Storage
Configure your file storage structure for document management and organization. This enables efficient access to documents and media across your platform.
4. Create Agents
Set up Agents with specific instructions and assign them the necessary tools and capabilities. Connect Agents to create expert networks for complex tasks that require specialized knowledge.
5. Develop Scenarios
Create Scenarios by defining prompts that will be triggered manually, on schedule, or via webhooks. Assign relevant Agents to these Scenarios to automate workflows.
6. Build User Interfaces
Use UI tools to create interactive experiences for data collection and presentation. This enables intuitive interaction with your AI-powered systems.
7. Execute and Monitor
Once your setup is complete, execute Scenarios and monitor the performance and interactions of your Agents and Applications. Track results and refine your system as needed.
By following these steps, you can effectively utilize Brightsy AI to create intelligent systems that automate and enhance your enterprise data workflows.