Understanding and Managing Assistants in Gen AI Builder v1.2
What are Assistants
Assistants in Gen AI Builder are AI-powered agents that perform tasks, interact with users, access knowledge, and adhere to predefined behavioral guidelines. They are the central building block for creating interactive, AI-driven applications within the AI Factory ecosystem.
In short: Assistants turn your Knowledge Bases, Rulesets, and Tools into useful, conversational AI experiences.
For a deep dive, see: Assistants explained.
Why use Assistants
Assistants enable you to:
- Build interactive AI applications quickly and reliably.
- Ground AI responses in your organization’s data (via Knowledge Bases).
- Enforce behavioral guidelines through Rulesets.
- Enable advanced task execution through Tools.
- Maintain conversational memory to support multi-turn interactions.
Assistants are the primary interface between end users and your AI Factory content + capabilities.
When to use Assistants
- Whenever you want to expose AI capabilities to users — internal or external.
- When building:
- Support chatbots
- Internal knowledge agents
- Financial advisors
- Sales assistants
- Executive assistants
- Compliance checkers
- When building multi-modal Agents that combine:
- Natural language conversation
- Knowledge retrieval (RAG)
- Behavior control
- Task execution
How Assistants fit into Gen AI Builder
Typical Assistant flow:
User Input → Assistant → Retriever → Knowledge Bases → Retrieved Content → Rulesets → Behavioral Guidance → Tools → Action Execution (if configured) → Response Generation → User Output
At runtime:
- The Assistant receives user input.
- It retrieves relevant knowledge.
- It applies behavioral Rulesets.
- It uses Tools if needed.
- It generates a response via its selected LLM.
Assistants provide a unified layer over all these components.
Key features of Assistants
- LLM integration: Powered by your choice of Large Language Models (LLMs).
- Knowledge Base connectivity: Link one or more Knowledge Bases for Retrieval-Augmented Generation (RAG).
- Ruleset application: Enforce behavioral guidelines and tone.
- Tool usage: (If configured) enable Assistants to perform external actions.
- Conversation management: Maintain memory across interactions.
- Customization: Tailor Assistant behavior, knowledge access, and generation parameters.
Getting started
See Create an Assistant for a full step-by-step guide.
Typical workflow:
- Create an Assistant.
- Select LLM model.
- Add Knowledge Bases.
- Add Rulesets.
- Optionally configure Tools and Memory.
- Test and deploy the Assistant.