Databricks Launches Agent Bricks for Enterprise AI Agent Development

Databricks has introduced Agent Bricks, a new tool aimed at simplifying the creation of domain-specific AI agents for enterprise use. Now available in beta, Agent Bricks automates the entire agent development process—from evaluation and optimisation to deployment—based on high-level task descriptions and connected enterprise data.

The platform is designed to address frequent business needs such as structured data extraction, contextual knowledge assistance, custom text generation, and the orchestration of multi-agent systems. It offers an integrated, automated framework that eliminates the need for manually stitching together multiple tools or processes.

Agent Bricks incorporates research from Mosaic AI Research, enabling the automatic generation of synthetic data and task-aware benchmarks. These benchmarks are used to evaluate and optimise AI agents based on both performance and cost, significantly reducing the trial-and-error typically required in agent development.

According to Databricks, enterprises often struggle with scaling AI agents due to inconsistent quality, rising experimentation costs, and the fast pace of new AI model development. Agent Bricks aims to provide repeatable and domain-specific evaluations to help teams create reliable and cost-effective AI agents without requiring reskilling or specialised infrastructure.

Key Features and Use Cases of Agent Bricks

Agent Bricks follows a three-step automated process:

1. Automated Evaluation: Generates task-specific benchmarks and evaluators to assess agent quality.

2. Synthetic Data Generation: Creates training data modelled on a company’s domain-specific information.

3. Optimisation Search: Tests and fine-tunes agent performance across various quality-cost parameters.

This process produces production-ready AI agents optimised for industry-specific applications. Key use cases include:

• Information Extraction: Converts unstructured content (emails, PDFs, reports) into structured data fields. For example, retail teams can extract prices and product details from supplier documents of varying formats.

• Knowledge Assistance: Delivers accurate, cited responses using enterprise documents, improving technician access to operational content in sectors such as manufacturing.

• Multi-Agent Orchestration: Enables coordination of multiple agents for complex workflows. In financial services, this can include document analysis, compliance checks, and automated response generation.

• Custom LLM Agents: Supports text transformation tasks such as generating branded content, marketing copy, or customer-specific chat responses tailored for business needs.

By combining automation, synthetic data, and enterprise-grade controls, Databricks’ Agent Bricks is designed to reduce the time and complexity involved in building reliable AI agents, helping organisations integrate AI into operations more efficiently.

“Agent Bricks is a whole new way of building and deploying AI agents that can reason on your data,” said Ali Ghodsi, CEO and Co-founder of Databricks. “For the first time, businesses can go from idea to production-grade AI on their data with speed and confidence, with control over quality and cost tradeoffs. No manual tuning, no guesswork and all the security and governance Databricks has to offer. It’s the breakthrough that finally makes enterprise AI agents both practical and powerful.”

“With Agent Bricks, our teams were able to parse through more than 400,000 clinical trial documents and extract structured data points — without writing a single line of code. In just under 60 minutes, we had a working agent that can transform complex unstructured data usable for Analytics.” — Joseph Roemer, Head of Data & AI, Commercial IT, AstraZeneca.

“With Agent Bricks, we can quickly produce domain-specific AI agents for tasks like extracting insights from customer support calls—something that used to take weeks of manual review. It’s accelerated our AI capabilities across the enterprise, guiding us through quality improvements in the grounding loop and identifying lower-cost options that perform just as well.” — Chris Nishnick, Director of AI, Lippert.

“Agent Bricks enabled us to double our medical accuracy over standard commercial LLMs, while meeting Flo Health’s high internal standards for clinical accuracy, safety, privacy, and security. By leveraging Flo’s specialized health expertise and data, Agent Bricks uses synthetic data generation and custom evaluation techniques to deliver higher-quality results at a significantly lower cost. This enables us to scale personalised AI health support efficiently and safely, uniquely positioning Flo to advance women’s health for hundreds of millions of users.” - Roman Bugaev, CTO, Flo Health.

“Agent Bricks allowed us to build a cost-effective agent we could trust in production. With custom-tailored evaluation, we confidently developed an information extraction agent that parsed unstructured legislative calendars—saving 30 days of manual trial-and-error optimization.” — Ryan Jockers, Assistant Director of Reporting and Analytics at the North Dakota University System.

“With over 40,000 complex legal documents, we needed high precision from our internal 'Regulatory Chat Tool’. Agent Bricks significantly outperformed our original open-source implementation (built on LangChain) in both LLM-as-judge and human evaluation accuracy metrics.” — Joel Wasson, Manager Enterprise Data & Analytics, Hawaiian Electric.

Databricks Expands GenAI Capabilities with New Mosaic AI Features

At the Data + AI Summit, Databricks announced new additions to its Mosaic AI portfolio, further supporting enterprises in building and deploying production-grade generative AI applications. The updates include serverless GPU support and the release of MLflow 3.0, both designed to streamline model development, deployment, and management.

Databricks has introduced serverless GPU compute, allowing developers and data teams to fine-tune models, run deep learning workloads, and experiment with large language models (LLMs) without managing traditional GPU infrastructure.

This new feature offers scalable, on-demand access to high-performance computing resources, removing the need for manual provisioning and reducing operational overhead. It is aimed at accelerating AI development cycles while improving cost efficiency for enterprises working on compute-intensive tasks.

MLflow 3.0: Upgraded AI Lifecycle Management for GenAI

Databricks also launched MLflow 3.0, the latest version of its open-source platform for managing the AI development lifecycle. Tailored for generative AI workflows, MLflow 3.0 supports monitoring, tracing, and optimisation of AI agents across any deployment environment.

Key features include:

• Prompt Management
  

• Quality Metrics and Human Feedback Integration
  

• LLM-Based Evaluation
  

• Cross-Environment Comparison and Debugging Tools
  

Users can now integrate MLflow traces and evaluation results directly with their existing lakehouse data infrastructure, helping teams improve model accuracy using real production trace data. MLflow continues to see widespread adoption, with over 30 million downloads per month.

Combined with Agent Bricks, the newly announced low-code agent development framework, these Mosaic AI enhancements position Databricks as a comprehensive platform for end-to-end generative AI operations. From training and fine-tuning to evaluating and deploying AI agents, the platform aims to simplify complex workflows and reduce time to production for enterprise users.

Read More:

Zendesk Partner Program: A Simplified Guide for the IT Channel Ecosystem

Challenges of Indian System Integrators: Adapting to Survive and Thrive

Check Point's India Channel Strategy for Cybersecurity Growth

Navigating System Integration in the Digital Era: Overcoming the Challenges