Introduction
If you're working in the Prompt Library but struggling to understand which model to use for a given product, we've got you covered. Use the region-specific tables in this article as a guide based on the prompt's client type.
Built-in data protection
LivePerson works to ensure that your data remains secure and protected whenever generative AI capabilities are used, no matter which model powers them.
- Zero data retention: Your data is processed in-memory and is never stored by the model provider.
- No model training: Google is strictly prohibited from using your data to train their foundation models.
- Robust guardrails: All of our existing Generative AI security guardrails are fully supported to ensure sensitive data is protected.
For a comprehensive overview of our security architecture and data handling policies, we invite you to review our updated Generative AI Security & Privacy Whitepaper.
Important notes
- Model selection: LivePerson uses Google's Gemini models across our applications, selected through rigorous benchmarking for advanced, reliable Generative AI capabilities. As newer Gemini models become available, we evaluate and adopt them when they offer clear improvements. For some applications in the APAC region, no Gemini model that keeps your data in region is currently available, so we use Azure-hosted models instead. (Note that model selections can also be driven by the retirement of legacy models as determined by the third-party provider.)
- Move to a custom prompt early: During exploration, feel free to use the system prompts that are available by default. However, we strongly recommend that you move quickly to a custom prompt as soon as possible. This is because LivePerson can change the system prompts without notice, and this might alter the behavior of your solution. To move to a custom prompt, you can copy the LIvePerson prompt and then use the copy — with or without any edits. Alternatively, select another custom prompt from the Prompt Library.
- Stick to recommended models: While you may see other providers, subscriptions, and LLMs available in the UI, for optimal results, only use the recommended models below unless your LivePerson representative has guided you otherwise.
Models for North America region
| Client Type | Provider | Subscription | Recommended LLM |
|---|---|---|---|
| Auto Summarization* | summarization | Gemini-2.5-flash | |
| Copilot Rewrite | rewrite | Gemini-2.5-flash-lite | |
| Copilot Translate | translate | Gemini-2.5-flash-lite | |
| Conversation Assist - Messaging | kai | Gemini-2.5-flash-lite | |
| KnowledgeAI agent (messaging bot)** | kai | Gemini-2.5-flash-lite | |
| KnowledgeAI agent (voice bot)** | kai | Gemini-2.5-flash-lite | |
| Routing AI agent (messaging bot)** | ai-studio | Gemini-2.5-flash-lite | |
| Routing AI agent (voice bot)** | ai-studio | Gemini-2.5-flash-lite | |
| aiStudio Guided Routing v1 & v2*** | ai-studio | Gemini-2.5-flash | |
| aiStudio Data Collection v1 & v2*** | ai-studio | Gemini-2.5-flash |
* This product is Conversation Summaries.
** This product is Conversation Builder.
*** Doesn't use the Prompt Library UI; prompts are selected within aiStudio.
Models for EMEA region
| Client Type | Provider | Subscription | Recommended LLM |
|---|---|---|---|
| Auto Summarization* | summarization | Gemini-2.5-flash | |
| Copilot Rewrite | rewrite | Gemini-2.5-flash-lite | |
| Copilot Translate | translate | Gemini-2.5-flash-lite | |
| Conversation Assist - Messaging | kai | Gemini-2.5-flash-lite | |
| KnowledgeAI agent (messaging bot)** | kai | Gemini-2.5-flash-lite | |
| KnowledgeAI agent (voice bot)** | kai | Gemini-2.5-flash-lite | |
| Routing AI agent (messaging bot)** | ai-studio | Gemini-2.5-flash | |
| Routing AI agent (voice bot)** | ai-studio | Gemini-2.5-flash | |
| aiStudio Guided Routing v1 & v2*** | ai-studio | Gemini-2.5-flash | |
| aiStudio Data Collection v1 & v2*** | ai-studio | Gemini-2.5-flash |
* This product is Conversation Summaries.
** This product is Conversation Builder.
*** Doesn't use the Prompt Library UI; prompts are selected within aiStudio.
Models for APAC region
| Client Type | Provider | Subscription | Recommended LLM |
|---|---|---|---|
| Auto Summarization* | summarization | Gemini-2.5-flash | |
| Copilot Rewrite | Azure | lp-llm-ptu | gpt-4o-mini-2024-07-18 |
| Copilot Translate | translate | Gemini-2.5-flash-lite | |
| Conversation Assist - Messaging | Azure | lp-llm-ptu | gpt-4o-mini-2024-07-18 |
| KnowledgeAI agent (messaging bot)** | Azure | lp-llm-ptu | gpt-4o-mini-2024-07-18 |
| KnowledgeAI agent (voice bot)** | Azure | lp-llm-ptu | gpt-4o-mini-2024-07-18 |
| Routing AI agent (messaging bot)** | ai-studio | Gemini-2.5-flash | |
| Routing AI agent (voice bot)** | ai-studio | Gemini-2.5-flash | |
| aiStudio Guided Routing v1 & v2*** | ai-studio | Gemini-2.5-flash | |
| aiStudio Data Collection v1 & v2*** | ai-studio | Gemini-2.5-flash |
* This product is Conversation Summaries.
** This product is Conversation Builder.
*** Doesn't use the Prompt Library UI; prompts are selected within aiStudio.