Configure AI Services & Prompts (Automated Actions)

GermainUX AI Service can be customized in many ways. At a high level, here are the main options:

Overview

• AI features are prompt-driven.

• Prompts can vary by use case and data type.

• Some features can use offline/local models, but AI Explorer currently appears to depend more heavily on online models.

• Customers may need to approve online AI usage depending on privacy policies.

• Online models may be faster and stronger but have cost/privacy considerations.

• Offline models may be more privacy-friendly but require hardware and may have throughput/context-length limits.

• AI features can be globally enabled/disabled and may also have per-feature toggles.

LLM Selection

You can choose the LLM you want to run against the GermainUX dataset.

Path: GermainUX Workspace > System > AI > Prompts > Model field > Select LLM

Online or Offline

For most LLMs, you can choose to run them within the GermainUX AI Service (offline) or connect to the model hosted externally (online).

Path: GermainUX Workspace > System > AI > Prompts > Set Online/Offline for a prompt

AI Prompt Customization

You can customize any AI prompt to leverage preconfigured or custom KPIs.

Path: GermainUX Workspace > System > AI > Prompts > Select a prompt > Edit in right panel

Example of a Custom AI Prompt that is using several GermainUX KPIs to summarize a Session that was recorded by GermainUX:

Similarly, for a more agentic approach, you can leverage these AI features alongside GermainUX’s automation capabilities.