Nextcloud + AI: Automatically capture and process documents

In NextcloudAI can be used in two ways: via the built-in Nextcloud Assistant, which writes and summarizes texts, translates them, and answers questions about your own files with Context Chat – and via Automations, that detect, extract, and process new documents. The AI can run on a dedicated server; documents then do not leave the company.

The Nextcloud Assistant

Nextcloud significantly expanded its AI assistant in 2026 (Assistant 2.0 with Nextcloud Hub 26). It works directly in the interface – in files, mail, Talk, and texts. Typical functions:

• Writing and rewriting texts – for example email drafts or summaries.
• Summarizing long documents or conversation transcripts.
• Translating texts.
• Transcribe from audio recordings.
• Context Chat: Ask questions about your files and receive answers based on the documents.
• Perform tasks: such as creating calendar appointments or sending messages in Talk.

Especially important for companies: The Assistant can work with local AI models and, if desired, on a separate AI server with a graphics card. Alternatively, external services can be connected – in that case, their data protection terms apply.

Context Chat: Questions about your own documents

Context Chat is an example of RAG: Nextcloud indexes your files, searches for the relevant passages for a question, and lets the language model answer on that basis. Instead of searching through folders, employees can ask, for example: “What notice period is specified in the maintenance contract?” How such systems work and what matters is explained by Building a RAG system in the company.

Automatically capture and process documents

Beyond Nextcloud, documents are often connected to an automation:

1. A document lands in a Nextcloud folder — via scanner, email forwarding, or upload.
2. An automation (for example with n8n) detects the new file.
3. AI reads the content and relevant details, e.g. sender, date, amount, or case number.
4. The document is named, tagged, and moved to the correct folder.
5. The responsible person is informed or a case is created in the specialist system.

For a pure document workflow with invoices and contracts, a specialized document system is often the better choice – see Paperless-ngx + AI.

Data protection and permissions

• Local models keep documents in-house.
• Rights: Context Chat should only provide answers from files that the person asking has access to – check this in the settings.
• External AI services only with a data processing agreement and clear rules on which data may be sent there.
• Transparency: Employees should be able to see when an AI is responding.

Typical use cases in SMEs

• Administration: Summarize and classify incoming correspondence.
• Social economy: Search concepts and manuals by question.
• Trades: Quickly find datasheets and quotes again.
• Project work: Transcribe meeting recordings and derive tasks.

bettersorted relies for its customers on open, inspectable components – operated on servers in Germany and bundled in the automaisa Hub. We provide manufacturer-neutral advice, are a BAFA-registered consultant and authorized INQA coach. The Consulting and guided implementation can be subsidized through INQA-Coaching at 80%; the Funding Check shows the suitable path. For an initial, non-binding consultation: Contact.

Example: Administration with Context Chat

An illustrative scenario: An administrative department stores policies, templates, and protocols in Nextcloud. With Context Chat, employees ask: “What documents are required for a special use permit application?” The answer points to the relevant policy. A local model on a separate AI server ensures that documents do not leave the organization. New protocols are automatically summarized and announced in the team chat.

Checklist for getting started

1. Clean up folder structure and permissions – the AI only finds what is stored in a sensible way.
2. Set up an AI server with a graphics card or a German model service.
3. Test Context Chat with a workspace and real questions.
4. Check whether answers respect access permissions.
5. Train employees: ask good questions, verify sources.

Costs and funding

The costs for AI in Nextcloud consist of three parts: Setup (planning, installation, integration, testing), ongoing operation (server or hosting, updates, monitoring, data backup) and support for employees (training, rules, contact persons). We do not quote fixed prices because scope and starting conditions vary greatly. The software itself does not incur license costs for open-source solutions.

Funding is provided not for the technology, but for the consulting and guided implementation: Through INQA-Coaching the federal government covers 80% of coaching costs nationwide (up to €11,520, vouchers until 30/06/2028). A preliminary analysis can be funded through the BAFA consulting subsidysubsidize – with 80% in the new federal states, Lüneburg and Trier, otherwise 50%, for applications submitted by 31.12.2026. For investments, some federal states offer their own programs – see Funding opportunities for companies in Germany.

Frequently asked questions

Does Nextcloud have its own AI?

With the Nextcloud Assistant, Nextcloud offers an integrated AI interface. The actual language models can be operated locally or connected via external services.

What is Nextcloud Context Chat?

A feature of Nextcloud Assistant that answers questions based on your own files. It searches for relevant passages in the documents and lets the language model build on them.

Do my documents remain within the company when using Nextcloud AI?

Yes, if you use local models on your own hardware or hardware hosted in Germany. When connected external services are used, content is transmitted to them.

Does Nextcloud AI require a graphics card?

For good performance with local models, yes. Nextcloud supports GPU acceleration, for example for text generation, transcription, and Context Chat.

As of October 2026. Models, versions and licenses change quickly – check the current license with the provider before making a decision.