Go to the Channel Catalog to learn more about what plans have access to this channel. Sign into Quickbase, and then go to Pipelines > Channels.
Use the OpenAI channel to integrate with Open AI to unlock a wide range of benefits by enabling powerful language understanding, automation, and decision-support capabilities.
Steps
Type | Name | Description |
Action | Create chat completion | Generate a chat completion based on the provided prompt |
Connection
- API Key—your OpenAI API key (token)
- Organization/Project—optional parameters based on your OpenAI settings
Options
-
AI Model—selection of available models. Different models have different limits and performance.
-
Max Completion tokens—token limit for the API call. Default value 1000 tokens.
-
Service Tier—Optional. For customers subscribed to the scale tier service, enter the latency tier to use.
-
Temperature—Optional. Number between 0.0 and 2.0. Controls the randomness of the generated completions.
-
Top P—Optional. Number between 0.0 and 1.0. The cumulative probability of the top tokens to sample from.
-
Frequency Penalty—Optional. Number between -2.0 and 2.0. Controls the likelihood of the model repeating the same words or phrases.
-
Presence Penalty—Optional. Number between -2.0 and 2.0. Controls how much the model avoids using words or ideas that have already appeared in the generated text.
-
Response Format:
-
Text—Returns the response from OpenAI as a single text object. Default setting for how OpenAI returns the response.
-
JSON Object—Returns the response from OpenAI as a JSON object.
-
JSON Schema—Define the JSON schema of how OpenAI returns and structures the response. Note that some OpenAI models can contextually split the response in the object according to the key name and value type.
-
Example: From a received email, you can generate an executive summary and categorize the importance of the email (low, medium, high) based on the tone of the sender.
-
-
User Message—give instructions on what do you want the AI to do.
-
System Message—defines the behavior, context, and persona of the assistant.
Default value—“You are an assistant. Provide helpful and accurate responses to user inquiries and follow their instructions.”
Limits
You can only use the non-realtime text chat completion models provided by OpenAI.
Any other models that, for example, generate or analyze images cannot be used via the channel.
Example: Generate a weekly summary email for each open bid
Generate a weekly summary email for each open bid.
Get all currently open bids
Generate a summary for each bid with OpenAI
Send an email to stakeholders via Outlook