Testing - Eyebrow
Connect a diverse set of models—both third-party and UiPath-developed. Avoid vendor lock-in and keep pace with AI innovation.
UiPath, 2025
UiPath, 2025
UiPath, 2025
Combine specialized and foundational models—both third-party and UiPath-developed—in a secure, compliant, and orchestrated environment.
Specialized models
Quickly build and fine-tune models with UiPath specialized AI
Integrate foundational models
Use UiPath Connector Builder to integrate models such as ChatGPT and many others
Operate with trust, governance, and security
Control and manage GenAI features and third-party models with the UiPath AI Trust Layer
State-of-the-art AI models are foundational to the UiPath Platform. See how AI powers our core capabilities.
UiPath IXP is a multimodal data classification and extraction experience to unlock enterprise data. IXP turns all document and communications data into insight and action at scale. The ‘X’ in IXP not only represents extraction but also the growing number of diverse and data types supported.
Discover UiPath IXPAgent Builder is a secure, simple way to build, test, and launch AI agents and the agentic automations they’re executing—all from within our familiar, low-code, fully integrated Studio environment.
Discover UiPath Agent BuilderThe AI Trust Layer gives you a centralized, intuitive place to control and manage UiPath GenAI features—as well as third-party AI models—across all UiPath products. With comprehensive management and control capabilities, including an auditing and usage panel, you get the guardrails and tools you need for responsible AI.
Read about the AI Trust LayerDocument Understanding Out-of-the-Box models
Pre-trained, high-performing models for processing your most frequent document types.
See OOTB modelsDocPath LLM
A specialized LLM for structured, semi-structured, and unstructured documents.
Get CommPath user guideUiPath AI Trust Layer
Directly oversee and govern UiPath GenAI features across all UiPath products.
Read about the AI Trust LayerExplore our methodology and findings on its performance under real-world workloads.