AI in a social institution: how Stiftung andante works with AI workspaces, an AI search on SharePoint and workshops

At andante, employees work in very different functions and professional roles. Day-to-day work involves numerous written tasks: reports, emails, documentation and internal notes. At the same time, the professional and regulatory requirements for the quality and traceability of these documents are increasing.

Important information such as guidelines, checklists, concepts and templates is stored on SharePoint. It is there, but often hard to find at the crucial moment. And finally, many employees had heard of AI but were unsure what it can do, where its limits lie and how to use it sensibly.

This resulted in three needs: secure, shared access to AI, intelligent access to existing knowledge and targeted enablement of employees.

All employees work in protected AI workspaces on the Evoya platform. There they use AI for their daily tasks: drafting, structuring, summarising and translating texts. This lowers language barriers and makes internal and external communication clearer, more understandable and more efficient.

The existing SharePoint storage is connected to the AI search. The approved documents are transferred into a protected, AI-optimised knowledge base (vector database). New or changed files are synchronised continuously, so the content is always up to date.

Employees ask their question in natural language. The AI searches the stored documents, summarises the relevant content and refers to the sources used. This makes guidelines, quality documents, processes and checklists easy to find in everyday work, without changing the original storage in SharePoint.

So that AI does not remain just another new tool, Evoya AI supported the introduction with practical workshops.

Module 1: Basics, AI in practice and getting started with the platform. In around three hours, up to 30 participants learn the basics of AI in plain language. Together they develop an initial target picture: where does AI take work off their hands, where does it support them, where are the limits and risks? This is followed by an introduction to the Evoya platform.

Module 2: Prompt and agent engineering. A core team of “AI Champions” deepens its knowledge of the platform, learns to write good requests (prompts) and maps recurring tasks with suitable agents. This team then acts as an internal point of contact and multiplier.

Module 3: Use case mining and process integration. AI Champions and key people analyse core processes, assess where AI provides meaningful support and derive prioritised use cases. For selected priorities, first prototypes are created, such as specific agents or AI-supported workflows, as “lighthouse applications”.

In a social institution, much of the information is particularly sensitive. The AI search only uses approved documents and does not change the SharePoint storage. Data is processed with Swiss or European AI, in line with the Swiss FADP and the GDPR, and never used for AI training.

• Secure access for everyone: all employees use AI in protected workspaces.
• Knowledge that can be found: answers from guidelines, checklists and templates, with a reference to the source.
• Clearer communication: texts are created faster and are easier to understand.
• Internal core team: the AI Champions support the further use of AI at andante.
• First lighthouse applications: prioritised use cases and prototypes show the way for the next steps.

Read more: AI in nursing and care · Evoya Search · Workshop: opportunity mining