Gemini
Peeklogic AI Orchestrator’s Gemini integration makes it easy to interact with your data and workflows. Enter natural language queries and let Gemini generate insights, summaries, and guidance to help you work faster and smarter.
Generate images using Imagen 3
Generates images from text prompts.
Gemini fields:
- Name
- Aspect Ratio
- Sample Count: The number of images to generate. CRITICAL LIMITATION: Due to Salesforce platform constraints (Apex heap size limit), this value is currently restricted to 1. Setting it higher will cause the request to fail.
- Person Generation: Allow the model to generate images of people
Disclaimer on Person Generation: The Don’t Allow setting is a guideline to the model. Due to the nature of generative AI, the model may still generate images containing non-identifiable people, even when this setting is active.
Content Generation
Generates text content from prompts using Gemini AI.
Output Type: The Content Generation feature currently only supports text output. Selecting Image in the Response Modalities will result in an error.
Gemini fields:
- Name
- Model – Default: Gemini 2.5 Pro
Note on Deprecated Models: The model selection list may include deprecated and duplicate versions (e.g., Gemini 1.0 Pro Vision ). These models may be removed by Google at any time. It is highly recommended to select the most current, non-experimental models (e.g., Gemini 1.5 Pro Latest ).
- maxOutputTokens: The maximum number of tokens in the response.
- stopSequences: Sequences that stop generation.
- Temperature: The randomness of the model’s responses.
- topK: The number of tokens to consider for sampling.
- topP: The cumulative probability of tokens for sampling.
- Gemini_Text_safetySettings
- category: Defines the type of potentially harmful content to be filtered. Examples include: “Harassment,” “Hate speech,” “Sexually explicit,” and “Dangerous content.”
- threshold: Sets how strict the filtering should be for each category. For instance, you can set a threshold to block only content with a high probability of being harmful, or you can be more strict and block content even with a low probability.
