How ERGO Hestia reduced time-to-market with Lakebase and Mosaic AI Model Serving

2026-06-11 · Source: Databricks · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Data Science & Analytics · Depth: Intermediate, long

Summary

ERGO Hestia, a leading Polish insurance company, transformed its real-time pricing platform, which manages over 100 models and 1,000 variables. Facing challenges with scaling innovation velocity due to an external Azure PostgreSQL database and custom adapter layer, they partnered with Databricks. The new architecture unifies data and model serving within the Databricks Lakehouse using Lakebase for an Online Feature Store and Mosaic AI Model Serving Endpoints. This eliminated external systems, reducing model deployment time and centralizing governance via Unity Catalog for full traceability. The migration was incremental, starting with low-criticality endpoints, demonstrating 20ms latency and less than 5% CPU utilization under 40 requests per second. This approach accelerated model time-to-market, simplified operations, and enhanced compliance, now serving as a blueprint for enterprise-wide scaling.

Key takeaway

For AI Architects or MLOps Engineers building real-time pricing or decision engines, consolidating your data and model serving within a lakehouse platform like Databricks is crucial. This approach eliminates external system dependencies, significantly reducing operational overhead and accelerating model deployment. You will achieve faster time-to-market. This also ensures robust governance, enabling rapid responses to market changes and scaled innovation across your enterprise.

Key insights

Consolidating data and model serving within a unified lakehouse platform accelerates innovation, simplifies operations, and enhances governance for real-time applications.

Principles

• Unify data and model serving within a single platform.
• Incremental migration de-risks large-scale architectural changes.
• Centralized governance ensures traceability and compliance.

Method

ERGO Hestia adopted a staged migration: PoC (Weeks 1-3) with low-criticality endpoints, then production deployment (Weeks 3-6), followed by traffic splitting for real-world validation.

In practice

• Use Lakebase for unified relational data serving on Delta tables.
• Expose models via Model Serving Endpoints for direct API access.
• Implement CI/CD pipelines for automated deployments.

Topics

• Real-time Pricing
• Lakehouse Architecture
• Databricks Lakebase
• Mosaic AI Model Serving
• Unity Catalog
• Incremental Migration
• MLOps

Best for: MLOps Engineer, AI Architect, Director of AI/ML

Related on AIssential

• Why Databricks is Winning the Data Wars in 2026 — A Technical Breakdown — Databricks' Lakehouse architecture unifies data warehousing and data lake functionalities on open standards, driven by governance and AI needs.
• Unlock seamless and cost-effective marketing campaigns with Lakebase — Lakebase merges transactional databases with data lake economics for scalable, cost-effective marketing data serving.
• MCP Marketplace Brings Real-Time Intelligence to Agentic Applications — Agentic applications require real-time external intelligence to move beyond historical data and make informed, autonomous decisions.
• Designing a Modern Enterprise Lakehouse with Databricks, DABs, and Medallion Architecture — Platform scalability in enterprise lakehouses relies on reusable engineering patterns and consistent deployment strategies, not just individual technologies.
• Accenture partners with Databricks on scaling enterprise AI solutions — Accenture and Databricks partnered to help enterprises scale AI by integrating data and deploying advanced AI agents.
• Databricks is no longer about tuning knobs — Databricks prioritizes abstraction and automation over granular control to serve less technical users and accelerate business value.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Databricks.