Aionic Labs builds Time Series Language Models, a model class that takes both text and raw temporal signals as input and output, so a single system can perceive sensor data natively and reason about it in natural language. Existing time series foundation models can forecast but cannot explain, while language models can explain but cannot perceive continuous measurement streams; the TSLM approach closes that gap and reports order-of-magnitude gains in temporal reasoning on smaller, faster backbones than frontier general-purpose models. The first model, OpenTSLM, came out of an open research collaboration between ETH Zurich's Agentic Systems Lab, the Stanford Byers Center for Biodesign and Google Research, was accepted at ICML 2026, and scores 99.71 on the TSQA reasoning benchmark against 59.24 for GPT-4o. Founded in Zurich in 2025 by Dr. Robert Jakob, who directs the ETH Agentic Systems Lab, with Thomas Kaar, the company was selected by Germany's SPRIND breakthrough innovation agency for its Next Frontier AI programme, worth up to €26.5M across stages, and is running pilots in manufacturing, healthcare, energy, logistics, telecom and banking.