Lightly AG
Category: Computer Vision
Data-centric AI company providing a computer vision suite for data curation, model pretraining, fine-tuning, and edge deployment, helping ML teams build better vision systems through smarter data selection. Lightly AG was founded in 2018. The company is led by Matthias Heller. Based in Zurich, Switzerland. Team size: 11-50. Total funding raised: $10M. Latest round: Seed. Key investors include Wingman Ventures, Y Combinator, Founderful, SICTIC, Soma Capital, First Momentum Ventures, Swisspreneur Syndicate, Innosuisse, Gaingels, Unlock Advisors.
- Founded
- 2018
- Headquarters
- Zurich, Switzerland
- Team size
- 11-50
- Total funding
- $10M
Value proposition
Helps ML teams reduce labeling costs by up to 92%, improve model performance by 19%, and accelerate retraining cycles by 2.1x through intelligent data curation that identifies the most valuable 1% of data to label.
Products and solutions
LightlyStudio (data curation, labeling, QA, and dataset management platform), LightlyTrain (self-supervised vision model pretraining and fine-tuning framework), LightlyEdge (smart data selection SDK for edge devices), LightlyServices (AI training data services for LLMs, CV, and RL), LightlySSL (open-source self-supervised learning framework), LabelFormat (standardized data labeling format)
Unique value
All-in-one computer vision suite from data curation to edge deployment, with open-source core and self-supervised learning that tells teams which data matters most before they label it.
Target customer
ML teams in autonomous driving, medical imaging, visual inspection, robotics, defense, agriculture, retail, and security — from Fortune 100 enterprises to startups
Industries served
Autonomous Driving, Medical Imaging, Visual Inspection/Manufacturing, Robotics, Defense, Agriculture, Retail, Security, Government
Technology advantage
Self-supervised learning for data selection; open-source framework (3.7K+ GitHub stars); ETH Zurich spin-off research pedigree; Y Combinator S21 batch; supports AWS S3, Google Cloud, Azure storage integrations
How they differentiate
Unlike pure annotation platforms, Lightly focuses on data curation before labeling — using self-supervised learning to identify the most valuable data samples. Open-source core (LightlySSL) builds community trust and adoption. End-to-end suite covers curation, pretraining, fine-tuning, and edge deployment in one platform.
Main competitors
Scale AI, Encord, SuperAnnotate, Dataloop, iMerit, Labelbox, Roboflow
Key partnerships
Y Combinator (S21 batch), Innosuisse (Swiss innovation funding), ETH Zurich (spin-off origin)
Notable customers
Fortune 500 automotive companies (autonomous driving), medical imaging enterprises, visual inspection companies
Major milestones
2018: Founded as ETH Zurich/HSG spin-off (originally WhatToLabel), 2020: Pre-seed round (CHF 1M from First Momentum Ventures), 2021: Y Combinator S21 batch ($125K), 2022: $3M Seed round led by Wingman Ventures, 2024: Reached $3M revenue; selected for Innosuisse Swiss Accelerator, 2025: Launched LightlyEdge; featured helping European carmakers, 2026: Launched LightlyStudio; CVPR 2026; 3.9K+ GitHub stars
Growth metrics
100+ ML teams worldwide; 3.9K+ GitHub stars; $3M revenue (Dec 2024); 1,200+ Discord community members
Market positioning
Data-centric AI platform for computer vision, positioned as a smarter alternative to traditional data labeling companies by focusing on data selection quality over annotation volume. Trusted by Fortune 100 companies with 100+ ML teams worldwide.
Geographic focus
Global, with strong presence in Switzerland (HQ), US (San Francisco office), and Europe
About Matthias Heller
Ex-private equity, M&A, and consulting professional; BA from University of St. Gallen, MSc in Strategic Management from HEC Paris; Y Combinator alumni
Latest news about Lightly AG
More Computer Vision companies
Official website: https://lightly.ai