aiLights - Swiss Public Inference Utility
How do we move AI from hackathons into reliable public services?
In this session, Oleg Lavrovsky presents the Swiss Public Inference Utility (SPIU), a newly established cooperative advancing the mission of Apertus, Switzerland’s fully open and reproducible language model developed by the Swiss AI Initiative. SPIU focuses on building sustainable public AI infrastructure aligned with Swiss values of sovereignty, neutrality, and inclusivity.
The session shares concrete insights into SPIU’s current work. This includes market validation efforts, community-driven governance, and the development of practical evaluation tools such as benchmarks, cost projections, and guided assessment wizards. We will also look at how the “Try Apertus” experience is being refined to highlight real-world use cases, particularly in education and public consultation, where transparency and auditability are critical.
Beyond technical infrastructure, the session addresses SPIU’s long-term vision. Starting with small, low-cost proof-of-concepts and a cooperative growth model, the initiative aims to demonstrate value step by step. In the longer run, the discussion expands to legislative pathways and the idea of AI as a public service, exploring how Public AI could become a durable, government-backed utility serving society as a whole.
Speaker:
Oleg Lavrovsky
Oleg is a Data engineer and developer relations specialist with strong open-source technology expertise, having worked as a freelancer in a range of SMEs & startups since 2012, contributing to projects like Dribdat, Proxeus & Frictionless Data. Prior to joining the Swiss AI Initiative and the EPFL AI Centre, O lectured at HSLU in the Data Design + Art program, and supports the BFH in teaching and research collaborations.
📅 Date & Time: 22.01.2026, 9:00-10:00
⏱️ Duration: 60 minutes
🌐 Language: English
🗣️ Format: Seminar
📍 Location: Live on Zoom/YT
👥 Audience: AI practitioners
🔗 More information about aiLights: https://ailights.org/
We thank our partner Oertli-Stiftung for the support.
🔗 Weitere Informationen zu aiLights: https://ailights.org/