LLM
LLM 01
Why manage models in GStudio
Answers stay on message
The identity prompt sets the role and boundaries, and the knowledge base sets the scope. When nothing relevant is found, the agent gives a fixed reply instead of improvising.
Risky content stays out
Both input and output can be moderated, and a hit returns a preset reply. Personal data can be anonymized before it reaches the model.
Balance cost and quality
Pick a model per task, add a fallback and cap the response length, so simple questions don’t need the most expensive model.
How models are set up and governed in an agent
Which model, what to say, what to remember, what it can call and what it must not say: all set in the configuration.
Each agent picks its own model. Agent and FlowAgent can set a fallback model that takes over when the main model is unavailable.
Model settings at a glance
- 200Max turns of short-term memory
- 100Tools per agent
- 4Reasoning depths
- 3Output message types
Frequently asked questions
Available models come from the platform model catalog, and each agent picks its own. Agent and FlowAgent can also set a fallback model that takes over when the main model is unavailable.
Agent Building
Agents that search knowledge, query data and call tools until the request is resolved.
Perform advanced tasks and solve problems dynamically
Learn moreAudio Agent Building
Real-time voice agents for support, phone lines and speaking practice.
Real-time voice agents that answer calls and talk on your website
Learn moreWorkflow Building
Multi-step processes orchestrated visually and exposed as APIs.
No-Code AI Automation Platform
Learn moreKnowledge Base
Equip your AI Agent with extensive business knowledge
Learn moreCustom LLM
Enhance and maximize LLM capability with powerful tools
Learn moreNo-Code Agent Builder
Build AI Agents with intuitive interface
Learn more