LangFlow integration for building
and connecting custom workflows
- Investigating integration of LangFlow for
building visual workflows - Goal: empower admins to create and expose
custom logic through Private GPT
A Trusted Provider setup to
securely offload AI processing
across isolated systems
- Concept to support secure, multi-instance
deployments for enterprise tenants - Enables external AI containers to handle
model inference, while customer systems
manage data
MCP Integration to enable external
tool calling and contextual
extensions
- Evaluating support for connecting to external
MCP (Model Context Protocol) servers - Would allow the system to call tools or
services during a conversation (e.g., for
calculations, database queries, or API calls)
A Model Switcher to allow flexible
use of different LLMs
- Aims to allow system administrators to switch LLMs via a CLI command
- Would support a predefined list of officially
tested models - Compatible models would be listed in a
Model Hub, integrated into the License Portal - Models can be downloaded separately,
reducing installation mage size and speeding
up upgrades
