OpenAI Function Agent

This component in Nappai uses OpenAI’s advanced language models to help you generate text. Think of it as a helpful assistant that understands your requests and can use extra information to give you the best possible answers.

Relationship with OpenAI

This component directly interacts with OpenAI’s language models. It sends your requests to OpenAI, receives the response, and presents it to you in the Nappai dashboard. The quality of the output depends on the chosen OpenAI model and the information you provide.

Inputs

Inputs: This is where you type your request or question. The component will send this text to the OpenAI model for processing.
LLM: This setting lets you choose which OpenAI language model to use. Different models have different strengths and capabilities. You’ll need to select an appropriate language model from the available options.
Memory: This allows the component to “remember” previous interactions. This is useful for maintaining context in a conversation or series of related requests. Select a memory type from the available options.
Tools: This lets you add extra tools the OpenAI model can use to help answer your requests. For example, you might add a tool to search the web or access a specific database. Select tools from the available options.

Outputs

result: This is the text generated by the OpenAI model in response to your input. This output can be used in other parts of your Nappai workflow, such as sending an email, creating a document, or updating a database.

Usage Example

Let’s say you want to write a marketing email announcing a new product. You would:

In the Inputs field, type your request: “Write a marketing email announcing our new ‘Super Widget’ product. Highlight its key features: durability, ease of use, and affordability.”
Select an appropriate LLM (OpenAI language model).
Choose a Memory type (if needed for a multi-part email creation).
Optionally, add Tools (e.g., a tool to access your product database for specifications).
Run the component.
The generated email will appear in the result output. You can then copy and paste this email or use it as input for another Nappai component to send it.

Templates

This component is used in many Nappai templates, but its configuration will vary depending on the specific template. Check the individual template documentation for details.

Many components can work with the output of the OpenAI Function Agent. For example:

Email Sender: Send the generated email directly.
Document Creator: Use the generated text to create a document.
Database Updater: Update a database with information extracted from the generated text. (Specific components will depend on your Nappai setup.)

Refer to the Nappai documentation for a complete list of components and their functionalities.

Tips and Best Practices

Clear Instructions: Provide clear and concise instructions in the Inputs field to get the best results.
Experiment with LLMs: Different OpenAI models have different strengths. Experiment to find the best model for your needs.
Utilize Memory and Tools: Leverage memory and tools to improve the context and capabilities of the agent.
Iterative Refinement: You may need to refine your input or choose different tools to get the desired output.

Security Considerations

Data Sensitivity: Be mindful of the data you include in your input. Avoid including sensitive information unless the chosen OpenAI model and tools are configured for secure handling of such data. Consult the OpenAI and Nappai security documentation for best practices.
Prompt Engineering: Carefully craft your prompts to avoid unintended biases or outputs. Review the generated text before using it.