Central AI chat frontend for a non-profit GmbH
Open-source frontend (Open WebUI), model selection by cost and data protection, rollout and continuous development.
Challenge
Staff at a non-profit GmbH used scattered, partly private accounts with commercial AI providers – with unclear data-protection posture, no governance and rising licence costs. A central solution had to be GDPR-compliant, fit a tight budget and serve a range of use cases (research, drafting, internal knowledge work).
Approach
Built a central chat frontend on top of the open-source project Open WebUI. Assessed and selected available models along two axes: cost per token and data protection (EU hosting, data-processing agreements, training opt-out). Multiple providers integrated in parallel, with a roles-and-permissions concept and audit logging. Handover into regular operation including user enablement, plus continuous extension with new models and features.
Result
Staff access AI models through a single GDPR-compliant frontend – with transparent per-use-case costs and clear governance. Licence costs are significantly lower than with individual accounts; new models can be added without changing tools.