Questions and answers about PrivateGPT

PrivateGPT is a secure Danish AI platform for organisations. It runs inside your own Microsoft Azure tenant and gives employees access to advanced AI, specialised assistants, document knowledge and integrations — without data ever leaving your own IT environment.

PrivateGPT is a fit for organisations that want to use AI seriously and safely across employees, departments and processes. Especially those with real requirements around GDPR, data security, Microsoft integration, documentation and control over their own data.

No. PrivateGPT includes chat, but it's built as a full enterprise AI platform. You can create your own assistants, use your own documents and data sources, integrate with Outlook and SharePoint, build Skills, and connect the platform to internal systems.

PrivateGPT runs in your own Azure tenant and acts as your own AI platform. That gives you stronger data control, technical isolation, access management and the ability to build your own assistants, document sources, integrations and workflows. ChatGPT, Copilot and Claude are strong tools, but they run in the vendor's cloud and don't give you the same control over the whole platform.

Employees can write, rewrite, summarise, translate and analyse documents. They can find information, prepare for meetings, produce minutes and get help with emails, reports and decision briefs — all inside your own environment.

Yes. With document RAG, PrivateGPT can search your documents semantically and answer based on internal sources — SharePoint documents, uploaded files, websites, manuals, policies, reports, contracts or guidelines.

RAG (Retrieval-Augmented Generation) is the technology that lets a language model answer questions from your own documents — without your data leaving your environment, and with source citations you can verify. Rather than relying only on the model's training, the system retrieves the most relevant excerpts from your sources and generates the answer from them.

Yes. Where it's relevant, PrivateGPT can return answers with source references so the user can verify the answer and see exactly which material the AI built on.

Yes. PrivateGPT can compare documents, spot differences between versions, summarise changes and point to the relevant sections. Particularly useful for contracts, reports, policies and long documents.

Yes. With Artifacts and Skills, PrivateGPT can produce structured outputs like reports, tables, notes and presentation drafts. The AI doesn't just reply in chat — it produces something you can use directly in the work.

Yes. PrivateGPT supports meetings with recording, transcription and structured minutes. It can extract decisions, action points and deadlines so follow-up is easier.

PrivateGPT is deployed to your own Microsoft Azure tenant. Documents, indexes, configuration, chat history, storage and logs all live in your own Azure environment — not in a shared platform on our side.

By default, no. PrivateGPT is designed so data stays in your own Azure environment. When AI models are invoked, they run via the Azure resources and regions you have chosen.

No. Customer data is not used to train foundation models. Your prompts, documents and answers deliver functionality inside your own solution — not to improve general AI models.

PrivateGPT is built to support GDPR and compliance work. It runs in your own Azure tenant, gives you control over data location, supports access management via Entra ID, and can produce documentation you can use in DPIAs and internal risk assessments.

Yes. Access is managed via Microsoft Entra ID, security groups, environments, workspaces and assistants. Users only see the areas, assistants and data sources they have rights to.

Yes. PrivateGPT can fence off sensitive areas like HR, finance, legal or leadership into their own environments or workspaces — so content, assistants and data sources are governed to fit your security needs.

Yes. The platform can plug into your existing security and logging setup — Azure Monitor, Log Analytics or Sentinel — for better traceability and to make it easier for the security team to follow usage and events.

Yes. PrivateGPT integrates directly into Outlook as standard. Users can summarise long mail threads and get help drafting replies based on the thread context, the organisation's tone and relevant documents.

Yes. PrivateGPT connects to SharePoint so existing documents, pages and libraries can be used as the knowledge base. Access permissions are respected — users only get answers based on content they're allowed to see.

Yes. PrivateGPT is built for organisations already on Microsoft 365. It uses Entra ID for sign-in and access management, and can integrate with Microsoft data through relevant Microsoft APIs.

Yes. PrivateGPT can integrate with internal systems via APIs, MCP and Skills. Assistants can look up data, fetch context, produce drafts or support processes directly inside the systems the organisation already uses.

MCP stands for Model Context Protocol — a standardised way to connect AI assistants to external tools and data systems. With MCP, PrivateGPT assistants can work with data from internal systems in a structured way, still under the organisation's own security rules.

An assistant is a specialised AI agent with a specific role, specific sources and specific tools. An assistant could be built for HR, support, legal, audit, leadership or internal knowledge sharing.

Yes. You can create assistants that fit your departments, processes and language. Assistants can get their own instructions, access to selected documents, specific tools and placement in the right workspaces.

Skills are reusable workflows that describe how a specific task should be solved. A Skill can combine document RAG, model calls, APIs and other tools, so recurring tasks are handled consistently and efficiently.

Skills can be used for reports, analyses, document reviews, case preparation, comparisons, presentation drafts and other tasks where the organisation wants a fixed structure and a more standardised output.

PrivateGPT can be extended with tools and integrations that let it take actions — for example via APIs or Azure-based services. Which actions are enabled depends on your security requirements, governance and approval flow.

PrivateGPT is deployed as a standalone platform in the customer's Microsoft Azure environment. Typically we set up infrastructure, Entra ID sign-in, access groups, relevant data sources, assistants and the first use cases in close cooperation with the customer.

PrivateGPT is designed to get started fast — especially for organisations already on Microsoft Azure and Microsoft 365. The concrete deployment depends on your security requirements, data sources and target use cases.

Users sign in through Microsoft Entra ID, so PrivateGPT fits into the organisation's existing sign-in, security groups and access policies.

PrivateGPT is designed as one platform for the whole organisation. The commercial model depends on the agreement, but the product is built for broad rollout — not as a tool for a few individuals.

Yes. Many organisations start with a few prioritised use cases — for example document RAG, Outlook, meeting notes or an HR or support assistant — and then expand the platform with more assistants, integrations and Skills.

PrivateGPT is configured with advanced language models through your chosen setup. The platform is not locked to a single model provider, and model choice can be tuned to quality, price, security, data location and compliance requirements.

Yes. PrivateGPT can be set up with EU-hosted models and components, so the organisation can keep data processing inside the EU as standard where that's a requirement.

Yes. Different assistants and workflows can be configured with different models. Some tasks need high quality, others need speed, low price or specific compliance considerations.

PrivateGPT uses Azure-based AI resources as part of the customer's own cloud environment. The AI functionality is delivered through your own Azure resources under the security and data processing agreements that apply to your Microsoft environment.

A demo is typically a short online meeting where we show PrivateGPT in action and talk about your needs. We can focus on the product, security, documents, Outlook, SharePoint, assistants, Skills or specific use cases.

It depends on the goal. Often it makes sense to invite a mix of business, IT, compliance and one or two people from the departments that will use the solution first.

You don't need to prepare much. It helps if you have 1-3 concrete use cases or challenges you want to explore — for example document work, internal knowledge sharing, meetings, support or legal processes.

Book a meeting and we'll have a no-strings conversation about your needs, security requirements and possible first use cases. From there we can assess how PrivateGPT is best set up for you.