Databricks Conference: Agent Bricks and Lakebase Take Center Stage

Data
Data Architecture and Engineering
Data Warehousing
Share this content:

Brian Rowe

Vice President of Solutioning, Data & AI
+ Connect on Linkedin
Brian Rowe is an executive leader driving enterprise AI transformation strategies and solutions at Marlabs. He architected our proprietary Prompt Router platform, which provides secure enterprise AI access and advanced analytics capabilities.
Related Services:
Data
Data Architecture and Engineering
Data Warehousing
AI
Interactive AI

The recent Databricks Data + AI Summit drew over 22,000 attendees to Moscone Center. With a global audience of 65,000 across 150 countries, the conference emphasized Databricks’ commitment to democratizing data and AI, underscored by major product launches and groundbreaking announcements.

P
More of our team at the summit

While numerous innovative developments emerged, two products particularly stood out to me: Agent Bricks and Lakebase. Here’s a focused exploration of these game-changing tools.

Agent Bricks: Simplifying AI Agent Development

Agent Bricks, introduced in beta on June 11, 2025, addresses a persistent challenge — building and optimizing AI agents effectively with minimal manual effort.

Key Capabilities:

  • No-Code Agent Builder: Users define tasks in natural language (e.g., knowledge extraction, chatbot creation), then simply plug in their data — Agent Bricks automates the rest.
  • Auto-Evaluation and Synthetic Benchmarks: Generates evaluation sets and leverages LLM judges, bypassing the manual labeling bottleneck.
  • Auto-Optimization: Tunes prompts, selects optimal models, and performs fine-tuning automatically, balancing cost and quality seamlessly.
  • Continuous Improvement (ALHF): Incorporate agent learning from human feedback to enhance performance continuously.

Why It Matters:

Agent Bricks dramatically accelerates AI deployment. Enterprises can now create production-ready agents in days, significantly reducing time-to-market. Benchmarks demonstrate substantial cost reductions (up to 10x lower, according to Databricks) alongside enhanced accuracy compared to proprietary models.

Real-World Impact:

They boast companies like AstraZeneca, Experian, and Block already leverage Agent Bricks to build powerful AI applications, from structured information extraction to orchestrating multi-agent systems for compliance and customer engagement. How do I get on this early user list?

It looks like it is making sophisticated AI capabilities accessible and actionable for enterprises without heavy infrastructure overhead.

One area showcasing data intelligence at the summit

Lakebase: Real-Time Transactional Power in the Lakehouse

Another standout announcement was Lakebase, Databricks’ new serverless, Postgres-compatible OLTP database integrated directly into the lakehouse ecosystem.

Core Advantages:

  • Transactional + Analytical Integration: Lakebase merges real-time transactional capabilities with analytical workloads seamlessly within one platform.
  • Sub-10 ms Latency: Optimized for high-volume transactional workloads, making it ideal for AI-driven applications, recommendation engines, and operational dashboards.
  • Modern Developer Experience: Offers branching, instant cloning, and autoscaling compute — essentially a Git-like experience for database management.
  • Unified Governance: All data operations integrate effortlessly under Unity Catalog, streamlining security, compliance, and data protection.

Differentiating Lakebase from Delta Lake:

While Delta Lake remains ideal for analytical workloads involving batch processing, Lakebase focuses specifically on transactional, real-time operations — think this one can handle fast inserts and updates.

Why Lakebase Matters:

Lakebase bridges a critical gap by enabling real-time application development alongside robust analytical capabilities. Companies no longer require separate systems or costly integrations — one platform now supports both operational agility and deep analytics.

The lawn at the summit at Moscone Center

Conference Themes: Simplifying Complexity

Databricks’ overarching theme was clear: reduce enterprise data and AI complexity. Centralizing data management through open formats like Delta Lake and Iceberg, coupled with unified governance via Unity Catalog, positions organizations to thrive in the AI era. Databricks reinforced this strategy with stronger integrations across cloud platforms (Azure, AWS, GCP), emphasizing open-source contributions and community-driven growth.

Looking Ahead

The launches of Agent Bricks and Lakebase are significant steps toward a future where AI and data-driven decision-making become accessible, fast to market, and deeply integrated into enterprise workflows. With these tools, Databricks continues to empower businesses, reducing barriers to innovation, accelerating AI adoption, and ultimately driving strategic growth.

The Data + AI Summit 2025 had so much more going on as well. There are so many YouTube videos you watch if you wanted to see some sessions.

Here is a link: https://www.youtube.com/playlist?list=PLTPXxbhUt-YX2FtEjfzzuPDDXxqRI6rLm