Chatbot Features
Last updated
Was this helpful?
Last updated
Was this helpful?
Glik’s Agent Assistant can be deployed as a chatbot that leverages large language models (LLMs) to automate tasks, analyze data, and make decisions. This conversational interface makes it easy for businesses to integrate AI-driven assistance into their workflows.
Agent Assistant Templates: Glik provides pre-built Agent Assistant templates in the Studio section. These templates offer a quick start for common business tasks and can be added directly to your workspace.
Creating Custom Agent Assistants: Using the Glik environment, you can build a custom Agent Assistant tailored to specific tasks. The Visual Workflow Builder and other AI-centric tools in Glik help you define processes, integrate data sources, and customize logic without heavy coding.
Model Selection: Agent performance depends heavily on the selected LLM. We recommend GPT-4 or other high-capability models for more stable and accurate task completion.
Prompt Configuration: You can define the Agent Assistant’s objectives, workflow, and constraints in the Instructions field, ensuring clarity and better results.
Knowledge Base Tools: In the Context section, you can add internal or external knowledge bases. The Agent Assistant can query these for additional information.
Functional Tools: In the Tools section, you can enable functionalities like internet search, complex calculations, or image generation. Glik supports both built-in tools and custom tool integrations (e.g., via OpenAPI/Swagger or OpenAI Plugin standards).
Enhanced Capabilities: By orchestrating the right set of tools for the Agent Assistant, you allow it to execute code, retrieve proprietary data, and perform more complex tasks through step-by-step reasoning and tool usage. Simply mention the relevant tool by name in the conversation, and the Agent will automatically invoke it.
Glik offers two inference modes for Agent Assistant:
Function Calling – Recommended for models like GPT-3.5 or GPT-4 that support it, offering improved consistency.
ReAct – Used by models that do not support function calling, emulating a similar structured reasoning process.
You can also adjust the iteration limit in the Agent settings to control how many reasoning steps the Agent Assistant takes.
You can define an initial greeting or prompt that appears when users first interact with the Agent Assistant. This conversation opener highlights the Agent’s capabilities and provides example questions to guide user interaction.
Before publishing your Agent Assistant as an application, you can debug and preview its behavior. This step helps you validate task completion, troubleshoot issues, and refine any workflows or tool integrations.