ZBrain

The ZBrain piece in ZBrain Flow provides direct access to advanced AI capabilities and knowledge management features of ZBrain. This powerful component allows you to leverage AI models, search through knowledge bases, query applications, and run automated agents within your workflows. With ZBrain integration, you can enhance your automations with intelligent decision-making, natural language processing, information retrieval, and autonomous task execution - bringing AI-powered capabilities to every part of your workflow process.

How to Use ZBrain Piece in ZBrain Flow?

Step 1: Select ZBrain as Your Connection

• Click on the '+' button in the Flow and search for ZBrain.

• Select ZBrain.

• Decide on the action you need, then select it. ZBrain Flow provides several options:

  • Knowledge Base Search – Search and retrieve information from your knowledge bases.

  • Query App – Send queries to your ZBrain applications.

  • Ask AI Model – Directly interact with AI models.

  • Run Agent – Execute autonomous agents to perform complex tasks.

  • App Previous Conversations – Retrieve query sessions and conversation history for a specific app within ZBrain.

  • Update Knowledge Base – Modify the details of an existing knowledge base in ZBrain.

  • Update App – Edit key properties of an existing ZBrain app.

How to Search a Knowledge Base?

Step 1: Configure API Connection

• If you haven't connected ZBrain yet, click on the ‘API Key’ field, then select ‘Create connection.’

• In the popup window that appears:

  • Enter a ‘Connection Name’ to identify this connection.

  • Paste your ‘API Key’ from your ZBrain account (found in Settings > My Account).

  • Click ‘Save’ to create the connection.

• If already connected, you'll see your ZBrain account name with a ‘Reconnect’ option if needed.

Step 2: Select Knowledge Base

• From the ‘Knowledge Bases’ dropdown, select which knowledge base you want to search. You can choose from all knowledge bases available in your ZBrain account.

Step 3: Enter Search Query

• In the ‘Query’ field, enter the search terms or questions to look for in your knowledge base. This can be a direct question, keywords, or a specific phrase.

Step 4: Set Maximum Size

• Use the ‘Max Size (tokens)’ field to limit the amount of content returned. The default is 2000 tokens.

How to Query ZBrain Apps?

Step 1: Configure API Connection

• For connecting your ZBrain account, refer to Step 1 in the "How to Search Knowledge Base?" section.

Step 2: Select App

• From the ‘Apps’ dropdown, select which ZBrain application you want to query. This displays all apps available in your ZBrain account.

Step 3: Enter Query

• In the ‘Query’ field, enter the question or prompt you want to send to the selected app. Frame your query according to what the app is designed to answer.

Step 4: Specify Conversation ID (Optional)

• If you want to maintain context from previous interactions, enter a ‘Conversation ID’. Leave this blank for a new conversation without prior context.

How to Ask an AI Model?

Step 1: Configure API Connection

• For connecting your ZBrain account, refer to Step 1 in the "How to Search Knowledge Base?" section.

Step 2: Select AI Model

• From the ‘Model’ dropdown, choose which AI model you want to interact with. Different models may have different capabilities and specialties.

ZBrain supports a wide range of powerful LLMs to cater to diverse enterprise needs. Below is the complete list:

OpenAI Models

GPT-3.5 Series

Model	Key functionalities
GPT-3.5 Turbo	General-purpose conversational AI Good balance of performance and cost 4K token context window Suitable for chatbots, content generation, and simple queries
GPT-3.5 Turbo 16K	Extended 16K token context window Same capabilities as GPT-3.5 Turbo Better for processing longer documents
GPT-3.5 Turbo 1106	November 2023 update with improved instructions following Better JSON mode Reduced hallucinations 16K context window
GPT-3.5 Turbo 0125	January 2025 update Enhanced reasoning abilities Improved instruction following 16K context window

GPT-4 Series

Model	Key functionalities
GPT-4	Advanced reasoning capabilities Strong performance on complex tasks Better at understanding nuance 8K token context window Multi-modal capabilities (with vision)
GPT-4 0125 Preview	January 2025 update Improved reasoning and instruction following Enhanced actuality 128K token context window
GPT-4 1106 Preview	November 2023 update Better instruction following Improved JSON mode 128K token context window
GPT-4 0614	June 2024 update Reduced hallucinations Better system message understanding 8K token context window
GPT-4 Turbo	More cost-effective than standard GPT-4 Faster response times Knowledge updated through April 2023 128K token context window
GPT-4 Turbo Preview	Preview version with the latest improvements Experimental features 128K token context window
GPT-4o	Optimized version with near GPT-4 performance Significantly faster response times Cost-effective 128K token context window Multi-modal capabilities
GPT-4o Mini	Smaller, more efficient version of GPT-4o Balance of performance and cost 128K token context window
ChatGPT-4o Latest	Consumer-facing implementation of GPT-4o Optimized for conversational use cases Includes the latest model updates

GPT-4.1 Series

Model	Key functionalities
GPT-4.1	Next generation full-size model<br>• Advanced reasoning and coding abilities<br>• Improved factuality and reduced hallucinations<br>• 128K context window
GPT-4.1 Mini	Smaller version of GPT-4.1<br>• Better performance-to-cost ratio<br>• Suitable for most enterprise applications<br>• 128K context window
GPT-4.1 Nano	Highly efficient, compact model<br>• Fast inference speeds<br>• Good for deployment in resource-constrained environments<br>• 32K context window

GPT-4.5 Series

Model	Key functionalities
GPT-4.5 Preview	Experimental preview of next-generation capabilities Advanced reasoning and planning Enhanced creative abilities 256K context window

Anthropic Models

Model	Key functionalities
Claude 3 Haiku	Fastest and most compact Claude 3 model Efficient for high-volume applications Good balance of speed and intelligence 200K token context window
Claude 3.5 Sonnet	Mid-range model with advanced capabilities Strong reasoning and instruction-following Excellent document analysis and summarization 200K token context window

Meta Models

Model	Key Functionalities
Llama 3-8B-Instruct	Open-weight 8 billion parameter model Good performance for model size 8K context window Suitable for deployment on edge devices
Llama 3-70B-Instruct	Large 70 billion parameter model Strong performance across tasks 8K context window Good for complex reasoning tasks
Meta/Llama 3-2-3B-Instruct-v1.0	Compact 2.3B parameter model Optimized for efficiency Good performance for size Suitable for mobile and edge applications

Google Models (Gemini)

Model	Key functionalities
Gemini 1.5 Pro	Advanced multimodal reasoning Strong performance across text, code, and vision tasks 1 million token context window Excellent for complex multi-step tasks
Gemini 1.5 Flash	Faster, more efficient version of Gemini 1.5 Good performance-to-cost ratio 1 million token context window
Gemini 2.0 Flash	The latest generation efficient model Improved reasoning and instruction following Enhanced multimodal capabilities 1 million token context window
Gemini 2.0 Flash 001	Updated version of Gemini 2.0 Flash Improved performance and reliability 1 million token context window
Gemini 2.0 Flash Exp	Experimental version with the latest features Advanced capabilities being tested 1 million token context window
Gemini 2.5 Pro Preview 03-25	Preview of next-generation capabilities Enhanced reasoning and planning Superior multimodal understanding 2 million token context window
Gemini 2.5 Pro Exp 03-25 Free	Experimental free version Similar capabilities to the preview version 2 million token context window

Mistral AI Models

Model	Key functionalities
Mistral Large	High-performance generalist model Excellent reasoning capabilities Strong at following complex instructions 32K token context window
Mistral Large 2411	November 2024 update Improved factuality and reasoning Enhanced instruction following 32K token context window
Pixtral Large 2411	Multimodal version with vision capabilities Strong image understanding and reasoning November 2024 update 32K token context window

Anthropic O Series

Model	Key functionalities
O1	Flagship model with strong reasoning Excellent at complex problem-solving Superior instruction following 128K token context window
O1 Preview	Preview version with the latest capabilities Experimental features 128K token context window
O1 Mini	Smaller, more efficient version Good balance of performance and cost 32K token context window
O3 Mini	Next generation compact model Advanced capabilities in a smaller package Improved reasoning and problem-solving 32K token context window

Step 3: Set System Instructions

• In the ‘System Instructions’ field, define the AI's behavior and context. Default is "You are a helpful assistant" but you can customize this for specific roles.

Step 4: Enter Your Prompt

• In the ‘Prompt’ field, enter the question, instruction, or content for the AI to respond to.

Step 5: Add Images (Optional)

• If your model supports image analysis, you can add ‘Image URLs’ by clicking "Add Item".

Step 6: Adjust Model Parameters

• Temperature: Control randomness (0-2). Lower values for more deterministic responses.

• Maximum Tokens: Set the length limit for the generated response.

• Response format: Choose between "Text" or other available formats.

• Top P: Adjust nucleus sampling parameter (alternative to temperature).

• Frequency penalty: Control repetition of phrases (-2 to 2).

• Presence penalty: Control topic diversity (-2 to 2).

• Messages: Set the number of message exchanges to include.

How to Run an Agent?

Step 1: Configure API Connection

• For connecting your ZBrain account, refer to Step 1 in the "How to Search Knowledge Base?" section.

Step 2: Select Agent

• From the ‘Agents’ dropdown, select which ZBrain agent you want to execute. This displays all agents available in your ZBrain account.

Step 3: Specify URL (Optional)

• If your agent needs to interact with a specific web resource, enter the URL. Leave blank if not required for your agent's operation.

Step 4: Provide Input

• In the ‘Input’ field, enter any data or parameters the agent needs to perform its task. The format will depend on what your specific agent expects.

How to Fetch Previous Conversations of an App

Step 1: Configure API Connection

• Enter the API key of an existing ZBrain connection. If you have not created one, refer to Step 1 in the "How to Search Knowledge Base?" section for instructions.

Step 2: Select App

• Choose the desired app from the dropdown. This will load the relevant app data from your ZBrain account.

Step 3: Select Session

• Select the session you would like to view. This displays previous query sessions associated with the selected app.

Step 4: Enter Limit

• Specify the number of recent conversations you want to retrieve by entering a limit. This helps narrow down the results based on your requirement.

How to Update a Knowledge Base

Step 1: Configure API Connection

• Enter the API key of an existing ZBrain connection. If you don’t have one, refer to Step 1 in the "How to Search Knowledge Base?" section to create a new connection.

Step 2: Select Knowledge Base

• From the dropdown, choose the specific knowledge base you want to update.

Step 3: Enter Title

• Provide the new title for the knowledge base entry you are updating. This will help identify the entry within the knowledge base.

Step 4: Enter Content

• In the content field, input the new or updated content for the knowledge base entry. This will overwrite any existing content associated with your selected knowledge base.

How to Update an App

Step 1: Configure API Connection

• Enter the API key of an existing ZBrain connection. If you don’t have one, refer to Step 1 in the "How to Search Knowledge Base?" section to create a new connection.

Step 2: Select App

• Choose the app you want to update from the available list in your ZBrain account.

Step 3: Enter App Name

• Provide the new name for your app or modify the existing name if needed.

Step 4: Enter Description

• Update the app description to reflect any changes or provide additional details.

Step 5: Select the Model

• Choose the model that your app will utilize. Select from the available options based on your app’s requirements.

Step 6: Enter the Temperature

• Input the temperature value for the app's model settings. The temperature controls the randomness of the model's responses (higher values make the output more random).

Step 7: Enter the Context Max Token

• Set the maximum token limit for the app’s input context. This defines the amount of data the model can consider while generating responses.

Step 8: Enter the Response Max Token

• Specify the maximum token limit for the app's output response. This determines the length of the model's generated output.