LLM Models

The LLM Models feature powers the intelligence behind workflows by enabling seamless integration with large language models. It allows workflows to understand input, generate responses, and perform complex reasoning tasks in real time.

This feature acts as the decision-making engine of your automation, transforming prompts into meaningful outputs while managing how models are configured, executed, and optimized. It supports both hosted and self-managed models, providing flexibility in how AI capabilities are deployed and scaled across workflows.

By using LLM Models, workflows can dynamically process context-rich input and generate outputs that drive downstream actions across agents, automations, and applications.

LLM Usage Modes

All LLM components support two primary usage modes:

1. Language Model Mode: In this mode, the model acts as an intelligence layer for AI agents. It provides reasoning, context, and decision support, enabling agents to interpret inputs, make decisions, and execute actions within workflows.

2. Model Response Mode: In this mode, the component directly processes the provided input and returns an AI-generated response. It is optimized for tasks such as text generation, summarization, and analysis, without requiring agent integration.

Core Capabilities

1. AI-Powered Execution: Enables workflows to perform intelligent tasks such as generation, reasoning, and conversation.

2. Flexible Model Integration: Connects to both cloud-based and self-hosted language models through standardized APIs.

3. Adaptive Response Generation: Generates context-aware outputs by combining system instructions, user input, and workflow data.

4. Controlled Model Behavior: Provides control over response behavior, execution timing, retries, and performance.

5. Seamless Workflow Integration: Embeds AI capabilities into agents, RAG pipelines, and automation processes.