Why StackOne for AI Agents?
StackOne simplifies building B2B SaaS agents by providing:- Unified API Access - Standardized interfaces across different SaaS providers
- AI-First Design - Tools designed specifically for AI agent development
- Multi-Tenant Support - Built-in handling of customer accounts and permissions
- Standardized Data Models - Consistent data structures across providers
Architecture Overview
StackOne Agent Tools are built on three core components:1. Unified API
Our standardized API layer that normalizes data across different platforms, ensuring more accurate and consistent results for your AI agents.2. Toolsets
Language-specific SDKs that wrap StackOne API endpoints into AI-friendly tools with native support for popular AI agent frameworks. Each toolset requires an API key to authenticate with StackOne API endpoints.3. Linked Accounts
Account identifiers that enable your agents to access the correct customer-specific tenant data securely and efficiently.Key Concepts
- Initialize a ToolSet: Create a
StackOneToolSet
instance to access unified business APIs - Filter Tools: Use glob patterns like
hris_*
to get specific tool categories - Execute Tools: Call business system APIs through unified tool interfaces
- Multi-Tenant: Each tool execution requires an
accountId
for customer isolation
TypeScript Framework Integrations
OpenAI Integration
Build agents with OpenAI Functions
Vercel AI SDK
Use with Vercel’s AI SDK for streaming and tool calls
TypeScript Core Features
Tool Filtering
Advanced glob pattern filtering for tool selection
Meta Tools
Dynamic tool discovery and execution (Beta)
Installation
Set up the TypeScript SDK
Basic Usage
Core concepts and fundamental usage patterns
TypeScript-Specific Features
- Type Safety: Full TypeScript support with comprehensive type definitions
- Framework Support: Native integrations with OpenAI, Vercel AI SDK, and more
- File Handling: Built-in support for file uploads and transformations
Next Steps
Resources
- GitHub Repository: Complete source code and examples
- Examples: OpenAI integration, meta tools, and file uploads
- Issues: Report bugs or request features
- NPM Package: Package details and version history