Public Roadmap for MAIA

We are constantly working to make MAIA the leading AI platform for knowledge management in organizations.

That is why we release relevant new features every few weeks. You will always find the most recent ones in the release notes.

🚧 In active development

Our goal for the next releases: MAIA should not just answer individual questions well, it should be the place where a team's work comes together. The knowledge about a project, the recurring procedures, and the systems your work already lives in.

📦 Features we are actively working on

• Projects - One shared place for everything that belongs to a single project: the files, the knowledge about it, and the chats around it. Files and project context are shared with your team, your chats stay private. Every chat in the project automatically knows the project files and the key decisions, so you do not have to explain them again each time. Decisions, requirements and lessons learned are captured in seconds, typed or dictated. Projects can be shared, handed over and archived without anything getting lost.

• Personal memory - MAIA remembers what matters to your work over the long run, instead of you repeating it in every new chat. You can see what is stored at any time in your settings, and you change or delete it simply by talking to MAIA. The memory belongs to your account, nobody else in the workspace can see it, and you can switch it off or delete it completely whenever you want.

• Custom instructions - Define once how MAIA should generally respond to you: form of address, language, format, recurring conventions. From then on this applies across all of your chats without you adding it every time.

• Answer effort and model selection - Instead of several technical toggles there will be a single choice: how fast or how thorough MAIA should be. Everyday questions become noticeably faster, while the full depth stays available for complex research across many sources. MAIA picks a suitable model automatically, or you select one yourself.

• Workflows - Capture a recurring task once as a sequence of steps and run it again and again, with the same reference documents and the same output format. Anyone running a workflow does not have to write a single prompt. The result is a normal chat and, if you want, a file as well. Workflows can be shared across the workspace so that a good working method does not stay in one person's head.

• File generation - MAIA will deliver results not only as text in the chat, but on request directly as a finished file to download: PDF, Word or Excel. An analysis, a report or a table then arrives right away in the format you pass on anyway.

• Integrations - MAIA will work directly with the systems your work already lives in. Connections are built on MCP, the open standard for connecting AI applications to external systems. Microsoft 365 and Atlassian come first, and a wide range of further MCPs will follow over the coming months. Your administrators decide which integrations are permitted in the workspace, and each person connects their own account. MAIA sees exactly what you are allowed to see yourself. Write actions always require your explicit confirmation before they are carried out.

• Deep Outlook integration - Emails and their attachments from personal and from shared mailboxes are synchronized with MAIA and analyzed in depth. This makes the part of your knowledge searchable and usable that so far only exists inside mailboxes: agreements, commitments, attached specifications and quotations.

🔒 Sovereignty

For many organizations it matters not only what an AI can do, but also where it runs and who operates it. We are working on this at two levels.

MAIA already runs exclusively in European data centers. We rely on server capacity from Microsoft Azure, Google Cloud, and Amazon Web Services, with a contractual guarantee that the service is hosted in Europe and that no data is transferred to the U.S.

We want to go further. Our goal is for MAIA to be able to run entirely on German systems in the future. To achieve this, we are relying on open-source models that make us independent of the major hyperscalers.

🔮 Within the next weeks

MAIA is already the smartest way for organizations to access their knowledge. We are not stopping there, and we are working on making sure the results reach you in the form you actually want to use them in.

🧱 Expansion: results, presentation and everyday usability

• Charts in chat - Figures from your documents or from the conversation directly as a finished chart, instead of a table you rebuild somewhere else. Ready to copy and download.

• Image generation - Create images directly in the chat and refine them in the conversation, without switching tools.

• Formula rendering - Mathematical formulas are typeset cleanly and readably in the chat instead of appearing as raw text.

• Many small improvements - More file formats, a copy button for code sections, drag and drop into newly created Data Collections, a collapsible sidebar, sharing chats and prompts within the workspace, a profile picture for your workspace, and a direct way to report a bug or request a feature from inside MAIA.