evidoria

← Back to browse

Good practice Imported

F13 — Baden-Württemberg's Sovereign AI Assistant for Civil Servants, Adopted Statewide by Thüringen

Germany · Stuttgart · See the Germany profile · See the Stuttgart profile

Evidence: Descriptive / self-reported Top 1% 93/100 · Ask Evidence Copilot about this practice

Baden-Württemberg built F13, the first AI assistant in productive use across a full German state administration, running on sovereign infrastructure; the open-source tool was then adopted by Thüringen under a one-for-all model instead of building its own system.

100,000+ users
Projected users (by end of 2026)

Details

Maturity
Scaling
Promoter
Staatsministerium Baden-Württemberg / InnoLab_bw (with Aleph Alpha, PD GmbH, BITBW, STACKIT)
Period
2024–2026
Keywords
digital government, generative AI, public administration, open source, civil service productivity

Context

Baden-Württemberg's state government, through its InnoLab_bw innovation unit, developed F13, a generative-AI assistant for civil servants, initially with Heidelberg AI firm Aleph Alpha, running on state-controlled infrastructure (BITBW's data centre and the STACKIT cloud) rather than external commercial clouds. It was presented publicly on 25 July 2024.

Objectives

F13 provides a chatbot, document-summarisation and research-in-administrative-documents tool for civil servants without relying on external commercial cloud providers; the state now maintains the full version independently of Aleph Alpha.

Activities

F13 was released as open-source software via the shared 'Open CoDE' platform for German public administration and built to be model-independent. Thüringen adopted the identical codebase under a 'one-for-all' approach, completing a pilot in late 2025 and planning full deployment through 2026 on its own on-premise cloud with a Thuringia-specific knowledge base.

Results

Adoption is projected to exceed 100,000 users across Baden-Württemberg by the end of 2026. Public reporting to date covers user-adoption and infrastructure milestones rather than task-level productivity metrics, and no independent evaluation of output quality or error rates has been published.

Implementation
Indicative cost
High (€500k–€5M) — Not publicly disclosed; involves state-funded infrastructure (BITBW data centre, STACKIT cloud) and an initial vendor partnership with Aleph Alpha.
Time to results
Medium (1–3 years) — Public launch July 2024 in Baden-Württemberg; Thüringen pilot completed late 2025 with full-scale deployment planned through 2026.
Staffing & skills
Baden-Württemberg state innovation unit (InnoLab_bw), External AI technical partner (Aleph Alpha) in the initial prototype phase, State IT infrastructure operators (BITBW, STACKIT, PD GmbH)

Conditions for success

  • Hosting on sovereign, state-controlled infrastructure instead of external commercial clouds
  • Releasing the codebase as open source (Open CoDE) so it is reusable and model-independent
  • A documented 'one-for-all' policy letting a second state adopt the same codebase rather than building its own

Common failure modes

  • No published task-level productivity metrics or independent evaluation of output quality/error rates

Where it fits

Governance type
state/regional government
Scale
statewide (two German federal states)
Income level
high-income

Commonly funded by

National / regional programmes Digital Europe Programme

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful