Toolhouse is a platform that enables developers to integrate tools into their projects to build powerful AI agents effortlessly. With Toolhouse, you can leverage pre-built tools and create applications that perform a multitude of useful tasks with minimal code.
You can start this journey with only three lines of code!
This repository explores examples of how you can use our pre-built tools to create applications for various use cases, such as customer service, blog writing, pet care, and more. It also demonstrates how to build an API that integrates the Toolhouse SDK with OpenAI's GPT models.
Feel free to check out our demo video to see Toolhouse in action.
git clone https://github.com/CharlesCreativeContent/toolhouse-example.git
cd toolhouse-example- Create a .env file in the root directory of the project and add your Toolhouse and OpenAI API keys:
TOOLHOUSE_API_KEY=your_toolhouse_api_key
OPENAI_API_KEY=your_openai_api_key- Install the required dependencies by running:
pip install -r requirements.txt- Once all dependencies are installed, you can start the server with:
python app.pyThe server will be running, and you can view the demo application at http://127.0.0.1:8000 and the API documentation at http://127.0.0.1:8000/docs
The FastAPI application includes a Swagger UI that allows you to test the various endpoints. The available endpoints are:
| Endpoint | Description |
|---|---|
| /pets | Answers pet owners using an veterinary clinics documents and can tell if the clinic is open |
| /blog | Generates a first draft of a research blog post including embedded links. |
| /customer | Provides customer service answers, adapting to open and closing hours. |
| Returns details about Twitter users and sends emails using hunter.io API. | |
| /test | A test route to check the integration of the Toolhouse SDK. |
- app.py: The main file that sets up the FastAPI application and integrates Toolhouse with OpenAI's GPT models.
- system_prompts.py: Contains pre-configured system prompts that shape the behavior of each endpoint.
- demo.html: Web page demonstrating the retrieval augmented generation (RAG) and search tools for a veterinary clinic.
Toolhouse acts as an orchestration layer between GPT models and various tools. By integrating the Toolhouse SDK, you enable AI models to perform actions based on decisions made during interactions. For instance, the AI can send emails or retrieve customer support information by adding a tool in Toolhouse without the need to update your code.
Building tools for AI agents can be time-consuming and complex. You need to:
- Define inputs and outputs.
- Implement robust error handling.
- Manage infrastructure to host the tools.
- Ensure effective communication with the model.
- Deal with different function-calling implementations across models.
Toolhouse simplifies this process by providing a platform where you can use tools written and maintained by developers like you. These tools work with any LLM that supports tool use. As most developers use more than one LLM provider in their deployments, Toolhouse helps you manage your functions across each model seamlessly.
- Ease of Integration: Start using powerful tools with minimal code changes.
- Community-Driven: Leverage tools developed by a community of developers.
- Cross-Compatibility: Works with multiple LLM providers and models.
- Scalability: Easily add or update tools without modifying your existing codebase.
Our growing community awaits you! We're looking forward to meeting you. While we focus on subjects relevant to Toolhouse, AI, or code, we're happy to chat about anything.