How to automate tasks with open source AI agent tools in Nextcloud Assistant

Post date

October 21, 2025

First things first: What is an AI agent?

AI agent tools help you automate routine tasks by using data and reason to perform authorized actions on your behalf, given a certain degree of autonomy.

Sounds pretty general? AI agents can be many things, depending on their complexity, knowledge of your data, or purpose. For example, an AI support agent can help you set up a help desk, automate easy help requests, and forward more demanding requests to your human team. An AI Sales assistant can answer product questions, schedule demos and meetings, and qualify leads.

Or an agent can be your AI copilot that finds data for you, helps you generate ideas and proofread your texts (or code), and performs tasks like sending emails and chat messages, creating files, and scheduling calendar appointments. Well, much like a personal assistant, which explains its name.

Under the hood, AI agents combine a Large Language Model (LLM) for an easy, human-like interface, API access to interact with your apps and data, and a layer of autonomy that lets them make data-driven decisions and carry out tasks as part of your AI workflows.

Proprietary AI agents: privacy trade-off for Big Tech convenience?

AI agents often require access to a wide range of user data: personal, financial, internal company documentation, and sensitive email communication. In proprietary services’ hands, even with robust security promises, this carries a great privacy risk. Using closed-source, third-party cloud-hosted models involves sharing a lot of sensitive data with a provider, which can lead to serious regulatory penalties and data breaches.

More often than not, agents built on proprietary models lack transparency regarding how they are trained, leading to biases and misinformation. In turn, given extensive permissions an AI agent may have, these bias can translate into operational roadblocks and errors.

However, private users and companies tend to trade off their privacy for the simple convenience that the robust external services can provide.

What can Nextcloud offer as an alternative, ensuring you can access the benefits of agentic AI without giving up your data?

How do open source AI agent tools work in Nextcloud Hub?

AI tools in Nextcloud are accessible via Nextcloud Assistant: your personal assistant that connects a variety of modern AI tools with your ecosystem and integrates deeply with some of the apps via specialized features.

Nextcloud Assistant screenshot Conext Chat

You can enable Nextcloud Assistant to work with your own data and files through context-aware tools, to ask questions about what’s happening in your workspace, or help find information in your complex file systems. For example, find a relevant chapter in documentation you struggle to find, or project to-dos your colleague sent in a Nextcloud Talk chat*.

To expand its abilities further, we introduced AI agent features that let Nextcloud Assistant perform actions with those apps and data and take part in automating complex, nuanced workflows.

Via chat interface, running quick agency-enabled commands is a very low-barrier experience, as you can simply use natural language. To create a calendar event, send a chat message in Nextcloud Talk, or check the weather for you.

Or do something more complex, like look up and chat with a customer, find out what they are looking for, and generate a template for a slide deck for a tailored presentation. All that, without risking any leaks of your sensitive conversations and documents — your security team can be sound asleep.

What you need to enable AI agent tools in Nextcloud Hub

Nextcloud Assistant relies on the Context Agent app as a back-end component for AI agent features. Make sure you install the app to enable the agent features in the AI Chat in Nextcloud Assistant. Don’t worry, that should be easy as pie. If you are a Nextcloud Enterprise customer, you can reach out to your account manager to receive support with setting it up upon request.

Hidden shortcuts: what AI agent features you can use in Nextcloud Assistant

Thanks to how well-integrated Nextcloud Hub is, with standards and under-the-hood connectivity provided by the open APIs, Nextcloud Assistant works with your data naturally across the Nextcloud Hub ecosystem, and in some cases beyond it.

Examples of supported AI agent features in Nextcloud Assistant:

Generate documents
List Nextcloud Talk conversations and send chat messages
List calendars and create calendar events
Check tasks in calendar
Create tasks and Deck cards
Search YouTube videos
List projects and create tasks in OpenProject
Search in DuckDuckGo
Plan routes on the map with different means of transport

Today, what you can do is no longer limited to just the Nextcloud Hub’s data and tools. You can easily work with integrated apps like OpenProject or external services like DuckDuckGo or online maps.

Our AI agent can also work with Context Chat and pull up data from your documentation, tasks, Nextcloud Collectives, Nextcloud Deck cards, and more. For example, you can ask Nextcloud Assistant, “According to Context Chat, what is the vacation request protocol?. It will then use Context Chat to find the data for you. With the Nextcloud Hub 25 Autumn release, Context Chat lets you search files using natural language and fetch links for them.

You can access the full list of currently available tools, including example prompts, on the documentation page.

And if by any chance you are an AI developer and would like to contribute to AI agent tools in Nextcloud Assistant, you are welcome to submit your PRs on GitHub!

Tired of typing? Use the voice input

Nextcloud Assistant now supports voice input and can answer using voice too, so you can ask it to perform all the actions it supports by simply talking to it. Perfect for when you are tired of typing or when you want to give a longer command.

Enable Nextcloud Assistant to work with your custom services via MCP

With the release of Nextcloud Hub 25 Autumn, Nextcloud Assistant now supports Model Context Protocol (MCP), which allows it to securely interact with external services. This way, it becomes a bridge between Nextcloud and other AI interfaces, such as AI support chat agents or even Smart Home devices.

You can configure MCP in your admin settings. Read about how to do it in the user documentation, which also includes technical requirements, model requirements, and essential steps.

Nextcloud Assistant as an AI agent: can it be trusted?

What matters is that Nextcloud Assistant is privacy-respecting by design. It is open source and you can use it with models you trust. And if you are unsure what to trust, we have an Ethical AI Rating to guide you around supported models.

Ethical AI in Nextcloud

We are all in on privacy and transparency. Nextcloud does not send your data to AI providers; instead, you can choose a model that fits your needs. Run it on your own server, use a trusted AI-as-a-Service provider, or use a service like OpenAI for non-critical tasks. In your settings, you can easily access the information about what AI tools are used in your Nextcloud and how. And importantly, none of them are enabled out of the box, so it is all entirely up to you!

In Nextcloud Hub 25 Autumn, we’ve worked on bringing you more control with some UI improvements. Nextcloud Assistant now lists the sources used for generating responses. In your settings, besides browsing the services used to power Nextcloud Assistant, you can turn them off and on as you wish. To increase the accuracy of the results and reduce costs, unconfigured integrations are turned off by default.

Run it on your terms: hosting the AI agent

Nextcloud is not an AI service provider, so the quality of the output and performance of the AI agent tools fully depend on your setup: the models you choose and the infrastructure you run your AI service on, in case you host your AI locally.

If you wish to have a managed AI to offload maintenance and load to a third party, you can explore the available AI as a Service providers. For example, you can set up your Assistant with a paid provider like OpenAI, Replicate, or IBM Watsonx, or go for a local, sovereign option like IONOS, plusserver, Groqcloud, MistralAI, or Together AI. We, of course, always advocate for transparent, local, and open-source options.

With Nextcloud, you don’t need to give away your AI data to Big Tech to automate your personal work: create content, find data in your documents, send emails and chat messages, carry out other operational routines, set up sales and customer support processes.

What else is new in Nextcloud Hub 25 Autumn

More agent tools are not the only updates Nextcloud Assistant received with the release of Nextcloud Hub Autumn. What else is new:

Ability to generate documents, spreadsheets, slides, and PDFs
File search with Context Chat using natural language
Live transcriptions in Nextcloud Talk
Multiple UI improvements
And more

Read more about all the new features in the Nextcloud Hub 25 Autumn release blog.

Nextcloud Hub 25 Autumn: Your digital workspace, ready in no time

Experience all the new features, including privacy-first AI automation, with Nextcloud Hub 25 Autumn. Start your instant trial of the latest release right away!