🚀 What’s New on Outmind

File upload is evolving with the arrival of drag & drop.

👉 You can now simply drop your files directly into the conversation, without going through the add icon 📎.

📌 Limits and supported formats

For reference, files remain limited to the conversation and are not added to the documents indexed in Outmind.

⇒ Enjoy a faster, more natural interaction.

You can now upload files directly from your mobile into a conversation with the Assistant.

This expands the upload feature already available on desktop: you can now use your documents wherever you are, without being tied to a device.

👉 It works the same way: add your files in the conversation, and they are immediately usable by the Assistant.

Following requests from many users, conversations with the Assistant are now kept with no deletion timeframe.

👉 You can find and reuse your exchanges at any time, even weeks or months later.

⇒ No need to rerun an analysis or lose context.

Two major improvements finally make tables truly usable.

Cell content now wraps automatically.

👉 No more unreadable tables stretching endlessly horizontally.

You can now open tables in a large, full-screen view for more comfortable reading.

👉 Ideal for exploring dense data or comparing information.

New options make it easier to control how your data is processed when configuring sources.

You can now manage and enable:

👉 These settings let you fine-tune Outmind’s behavior to match your needs.

⚡ What this changes

You can upload a file directly into a conversation with the Assistant, without indexing it in Outmind.

This is especially useful to work immediately on a document you’ve just received, without waiting for it to be indexed.

Add your file via the paperclip icon next to the send button, then select your document: it will be immediately usable in the conversation.

Note: this option is only available when activated by your administrator.

=> The file is not indexed and remains limited to the conversation

⚡What changes

Limits and supported formats

The limit is 20 documents per message sent to the Assistant.

The following formats are supported:

You can now select emails with the search tool, just like your files, and use them directly in a conversation with the Assistant.

Concretely, you can for example:

Simply select the relevant emails — alone or combined with documents — and start the discussion.

👉 The limit remains 20 items in total (emails + files).

⚡ What this changes

The Copy button now correctly preserves the formatting of the Assistant’s responses.

Headings, lists and tables are copied exactly as they appear in the interface, making them easier to reuse in other tools.

⚡ What this changes

We have introduced several improvements to make source management clearer, more consistent and easier to manage in Outmind.

These updates make it easier to configure connectors and better understand the status of your different data sources.

The configuration windows for local and cloud sources have been harmonized to provide a single, consistent interface, regardless of the source type.

Sources that have lost their connection were already identifiable thanks to a small crossed-out cloud icon in the sources list.

What’s new: you can now reconnect them directly in just a few clicks.

Simply click on the three dots to the right of the source, then select Reconnect to restore the connection.

When opening Outmind, a summary window also shows:

This makes it easier to stay in control of the status of your connections.

The Outmind assistant can now search the web when it’s relevant — for example, to retrieve up-to-date information or when you explicitly ask it to.

Results from the internet are integrated as a new data source, alongside your files and emails, with clear traceability of the websites used.

​Web search can be enabled or disabled at any time, on a per-conversation basis, and remains fully under admin control through configuration.
​

New assistants are now available in the assistant library, which is organized by categories (tenders, meetings, project follow-up, CVs, reference sheets, etc.).

Outmind has upgraded to GPT-5 and introduced an agentic workflow, delivering responses that are more reliable, autonomous, and context-aware — especially when analyzing complex files or tables.

GPT-5 brings more structured reasoning and a deeper understanding of relationships within data, while the new agentic system enables the assistant to plan, execute, and chain multiple actions autonomously — for example, think through a problem, analyze a file, then generate a coherent output.
​

What’s new

In short: GPT-5 thinks, acts, and verifies — producing expert-level answers.

Mobile-optimized Assistant version

Your field teams can now use the Outmind Assistant directly from their phone.

Please note:

You can now define a custom document scope for each AI assistant. It will answer only using the selected sources (folders, SharePoint, etc.).

✔️ Focus on a specific project or topic
✔️ Keep the scope consistent, even as documents change
✔️ Share specialized assistants with your team

A new Library section is now available in the Assistant menu. It lets you explore ready-to-use AI assistants created by Outmind, designed to help with specific tasks such as extraction, writing, verification, analysis, and more.

This update addresses a key need: providing efficient, out-of-the-box assistants to help users take immediate action without setup.
It unlocks new possibilities:

Available in the Assistant > Library menu for all users in teams where AI features are enabled.

Files located in the SiteAssets library of SharePoint sites are now included in search results.

The SiteAssets library is used to store resources associated with site pages, such as:

Outmind can now connect to iManage, the document management solution widely used by law firms and legal departments.

Thanks to this integration:

A faster, unified, and smarter way to find the right documents, wherever they are.

Outmind is now ISO 27001 certified — the international standard for information security management.

This certification demonstrates our commitment to protecting your data with the highest security standards, across all our services and infrastructure.

The search engine now automatically suggests reformulations or corrections when it detects an error in the query (such as a typo, incorrect use of an operator, etc.).

The AI now automatically reorders results to highlight the most relevant content based on context.

For admins: these features can be enabled per Team by going to Admin > Teams > Select a team > Features.

You can now ask the AI Assistant questions based solely on a specific set of documents.

This opens up many new possibilities, such as:

Following your feedback, it is now possible to adjust the width of the left-hand panel — which contains assistants and conversation history — in the Assistant view, just like in the Search view.

Based on your feedback, we’ve refined our relevance algorithm.
​

The most useful and contextually appropriate content now appears more effectively in the results, thanks to adjustments in the ranking engine’s weighting system.

Outmind continues to enhance its search engine performance and align its AI with the expectations of European companies.

The assistant will now rely on Azure OpenAI, a large language model hosted and operated by Microsoft in Sweden 🇸🇪.

This change addresses a key concern: ensuring data processing within Europe, while maintaining a high level of performance.

The assistant can now access the full (binary) file when needed to understand or generate a structured document (such as a presentation, Excel file, etc.).

This enables, for example:

You can now finely define the document scope the Outmind assistant relies on to provide answers.

From the Assistant view, a new menu allows you to:

✅ The assistant remains active within the defined scope

🔒 Data you don’t have access to is neither visible nor used

🧠 The scope can be adjusted before starting a conversation, but remains fixed once it has begun

You can now give the assistant custom instructions to tailor it to your role, preferred tone, and expectations.

Alias, role, response style… everything is configurable for more relevant and natural interactions.

The Assistant can now read documents in their full binary version, rather than relying solely on the indexed text version.

The result: more accurate answers, better adherence to the original structure, and a significant reduction in formatting or content interpretation errors.

The Assistant now automatically applies contextual filters based on your query — such as file format, date, source type (SharePoint, server, etc.), or chronological sorting.

You can now clearly see which filters and sources were used to refine the response.

Access your recent conversations with the Assistant, stored for 30 days to help you easily follow up and resume your exchanges.

The AI Assistant can now generate files directly from your conversations, thanks to the integration of the code interpreter.

Whether it’s tables, reports, exports, or visualizations — you can get ready-to-use files with just one click.

The Assistant now integrates DALL·E, OpenAI’s image generation tool.

You can create visuals on the fly from your conversations — illustrations, diagrams, icons, creative ideas — all directly within the interface.

Outlook users can now search within their own folder hierarchy — including inbox, subfolders, sent items, and more.

This allows for more precise targeting of the right emails, based on your personal organization.

You can now save custom prompts for yourself or your team.

Whether it's to automate recurring requests or make AI usage easier, this feature helps you save time and structure interactions.

✅ No more blank page syndrome
✅ Ready-to-use job-specific questions
✅ Better support for non-AI experts

Prompts are directly integrated into the DocChat conversational interface for quick, contextual access.