Databricks unveils Agent Bricks to streamline enterprise AI agents

Databricks has launched Agent Bricks, an automated product designed to allow enterprises to build and deploy AI agents that are customised for their data and operational needs.

Agent Bricks works by taking a high-level task description from users, connecting it with enterprise data, and then handling all additional tasks in the agent-building process. This includes generating synthetic data, running performance benchmarks and optimising the resulting AI agent.

The product is built using research from Mosaic AI and is currently available in Beta. It is aimed primarily at common business needs, such as knowledge assistance, information extraction, and the orchestration of multiple AI agents working together. Databricks has included built-in governance and enterprise controls intended to enable teams to implement Agent Bricks without assembling different components from multiple vendors.

Features and automation

Agent Bricks relies on synthetic data generation and automated evaluation to streamline the tuning of AI agents. According to Databricks, the workflow starts with automatic creation of task-specific assessments and large language model (LLM) judges to measure output quality. Synthetic data that matches the user's domain is then created to train and test the agent. Various optimisation techniques are applied automatically.

At the conclusion of this process, users select the iteration of the AI agent that reflects their chosen balance of quality and operational cost. Agent Bricks is designed to produce a domain-specific agent ready for use in business environments.

Industry use cases

The company outlined a series of use cases for Agent Bricks across different industries. In information extraction, the agent can process documents such as emails and PDFs, converting them into structured fields for easier analysis. For retail businesses, this means the ability to automate extraction of product information from supplier documents, regardless of formatting.

Other examples include the use of knowledge assistant agents in manufacturing, helping technicians quickly find answers in technical manuals, and enabling multi-agent orchestration in financial services to manage tasks like intent detection and compliance checks. Marketing teams can also use custom language model agents to generate content aligned with their brand's standards.

Addressing evaluation and scalability

Databricks states that Agent Bricks is a response to key barriers in deploying production-ready AI agents, especially the challenges of objectively evaluating new models for quality and cost—tasks that traditionally require manual processes and significant expertise. The automation of evaluation and data generation aims to make it possible to scale AI agent deployment without reskilling or expanding teams.

"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 own 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."

Customer feedback

Several Databricks customers have provided early feedback. Joseph Roemer, Head of Data & AI, Commercial IT, AstraZeneca, said, "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."

Chris Nishnick, Director of AI, Lippert, commented, "With Agent Bricks, we can quickly productionise 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."

Roman Bugaev, CTO, Flo Health, added, "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 specialised 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."

Ryan Jockers, Assistant Director of Reporting and Analytics at the North Dakota University System, said, "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 optimisation."

Joel Wasson, Manager Enterprise Data & Analytics, Hawaiian Electric, noted, "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."

Further AI platform releases

The launch of Agent Bricks is accompanied by additional tools. Databricks now provides serverless GPU support, which is intended to allow customers to fine-tune models or run deep learning workloads without having to manage underlying hardware. This provides on-demand and scalable access to computing resources for AI development and deployment.

Databricks has also released MLflow 3.0, the newest version of its open-source AI development framework. MLflow 3.0 is intended to help teams monitor, trace, and optimise AI agents across different environments, with integrated support for prompt management and evaluation. MLflow provides compatibility with existing data lakehouse architectures and continues to see substantial monthly usage figures.