Sherpa.ai

Privacy-first federated learning platform for collaborative AI training without sharing raw data

Insurtech Infrastructure Growth Series A-backed

Overview

Sherpa.ai is a SaaS platform specializing in privacy-preserving artificial intelligence, enabling organizations to train collaborative AI models without sharing raw data. Founded in 2012 by Xabi Uribe-Etxebarria, the company provides federated learning infrastructure that allows multiple parties to improve machine learning models while keeping sensitive data behind their own firewalls. The platform integrates federated learning with homomorphic encryption, secure multiparty computation, differential privacy, and proprietary technologies to support distributed AI training at enterprise scale.

The core offering is a SaaS platform that delivers federated learning capabilities through APIs and integrations, supporting both traditional machine learning and modern large language model fine-tuning. Organizations retain full data sovereignty while participating in collaborative model improvements—only encrypted model updates are shared between parties, never raw data. The platform supports edge deployment with on-device inference capabilities, ensuring data governance and regulatory compliance across distributed environments.

Sherpa.ai serves diverse industries including insurance, healthcare, financial services, telecommunications, manufacturing, aerospace and defense, and generative AI. Notable customers include NIH, KPMG, Zurich Insurance, Telefónica, Prosegur, Orange, Indra, Laboral Kutxa, and Osakidetza. The company has secured strategic partnerships with Telefónica Tech and KPMG to deliver federated learning solutions across their customer bases.

Products & Services

Federated Learning Platform (SaaS)

Privacy-preserving AI model training that enables distributed organizations to build and improve machine learning models on their own data without centralizing sensitive information.

Key Features

Federated learning architecture for collaborative ML training
Homomorphic encryption for computation on encrypted data
Secure multiparty computation (SMPC) for distributed calculations
Differential privacy for statistical privacy guarantees
Proprietary Blind Learning technology
Support for traditional ML and LLM fine-tuning
Edge deployment with on-device inference
API integration for seamless model training and inference

Target Users: Carriers, reinsurers, financial institutions, healthcare organizations, telecommunications companies, and other enterprises handling sensitive data

At a Glance

Founded: 2012
Headquarters: Erandio, Bilbao, Spain
Employees: 11-50
Funding: Series A-backed

Category & Focus

Category: Insurtech Infrastructure
Subcategories: Privacy-enhancing AI Federated Learning Platform Enterprise AI Infrastructure
Insurance Verticals: Reinsurance P&C Commercial Health
Target Customers: Carriers, Reinsurers, Enterprise Organizations

Customers

NIH (National Institutes of Health)
KPMG
Zurich Insurance
Telefónica (Spanish telecommunications major)
Prosegur (security services)
Orange (European telecom)
Indra (Spanish defense/IT)
Laboral Kutxa (financial services)
Osakidetza (Basque Country health service)
Telefónica Tech — partnership to deliver federated learning SaaS and professional services for analytics and AI across Spain
KPMG — agreement to offer AI capabilities while guaranteeing data privacy to their customers

Similar Companies

Last updated: 2026-06-17