VDF AI Chat icon

VDF AI Chat

VDF AI Chat is a private enterprise AI chat platform for governed retrieval, audit trails, and on-premise or air-gapped deployment.

VDF AI Chat

Private enterprise AI chat with governed retrieval

VDF AI Chat is a private enterprise AI chat platform built for organisations that need governed conversational AI inside their own perimeter. It combines chat, retrieval, orchestration, and logging so teams can use a ChatGPT-class assistant without routing prompts or documents to an external model provider.

The product centres on permission-aware private RAG: retrieved content is filtered by each user's access, and each response can be traced through an append-only audit record. It is designed to run on-premise, in a sovereign cloud region, or in an air-gapped enclave, with support for common enterprise knowledge sources and model endpoints.

Core capabilities

Private enterprise chat workspace

Runs chat, retrieval, embeddings, and conversation history inside the organisation's own security perimeter so prompts and documents do not need to leave the environment.

Permission-aware private RAG

Filters retrieval by source permissions and user access so answers are grounded in only the content the user is allowed to see.

Per-turn audit logging

Keeps an append-only record for each turn with identity, agent, model version, retrieved chunk IDs, and tool calls, with SIEM export support.

Flexible deployment modes

Supports deployment on-premise, in a sovereign cloud region, or in a fully air-gapped enclave, depending on the organisation's requirements.

Model choice with policy controls

Works with Qwen, DeepSeek, Mistral, Llama, and any OpenAI-compatible endpoint, with commercial APIs available only when policy allows.

Enterprise source connectivity

Connects to common enterprise sources including Confluence, Jira, GitHub, SharePoint/OneDrive, Google Drive, Notion, Fireflies.ai, and uploaded document sets.

Practical use cases

  • Private internal knowledge assistant

    Give employees a governed chat interface for searching internal documents and project knowledge without exposing content outside the company perimeter.

  • Audit-ready assistant usage

    Let compliance, security, and legal teams review a conversation trail that includes retrieved chunk IDs, model version, and tool usage for each turn.

  • Permission-aware enterprise search

    Use source ACLs to keep answers aligned with each employee's permissions when surfacing content from systems such as Jira, Confluence, SharePoint, or Google Drive.

  • Controlled deployment environments

    Run the assistant in a sovereign region or air-gapped enclave when organisational policy requires stricter control over data residency and external connectivity.

  • Multi-source enterprise workspace

    Connect multiple knowledge sources and model endpoints in one workspace so teams can move from a question to a grounded answer without copying context between tools.

Pros and Cons

Pros

  • Runs within an organisation's own security perimeter, including on-premise and air-gapped deployments.
  • Uses permission-aware retrieval so answers respect source-level access controls.
  • Provides per-turn audit evidence with retrieval provenance and tool-call history.
  • Supports a broad set of enterprise content sources and model endpoints.
  • Includes policy-gated optional use of commercial APIs rather than requiring them.

Cons

  • The page provides limited detail on the exact setup process and administration workflow beyond a standard on-premise deployment estimate.
  • Some platform capabilities are described at a high level on the source page, so specific connector behavior and workflow depth are not fully documented here.

FAQ

How is VDF AI Chat deployed?

VDF AI Chat is deployed on-premise, in a sovereign cloud region, or in an air-gapped enclave. The source also describes it as a private enterprise AI chat platform with permission-aware retrieval and per-turn audit logging.

What sources can it search?

It supports connected enterprise sources such as Confluence, Jira, GitHub, SharePoint/OneDrive, Google Drive, Notion, Fireflies.ai, and uploaded document corpora. Retrieval is permission-aware and filtered by the user's access.

Which models does it support?

The page describes support for Qwen, DeepSeek, Mistral, Llama, and any OpenAI-compatible endpoint. It also notes that commercial APIs are optional and policy-gated.

Who is it for?

The product is designed for organisations that cannot send prompts or documents to an external model provider. It is positioned as a governed conversational workspace with audit evidence for each turn.

How long does deployment take?

The source gives an approximate time to a governed pilot of about two weeks for a standard on-premise deployment.

Quick Facts

Category
Private enterprise AI chat / governed conversational AI platform
Deployment
On-premise, sovereign cloud region, or air-gapped enclave
Models
Qwen, DeepSeek, Mistral, Llama, and OpenAI-compatible endpoints
Connected sources
Confluence, Jira, GitHub, SharePoint/OneDrive, Google Drive, Notion, Fireflies.ai, uploaded documents
Commercial model policy
Optional and policy-gated
Source domain
vdf.ai