Why accesses plans to fix it

Many Enterprise Ai Agent Development Efforts Never Make it to Production and It’s Not Because The Technology Isnt Ready. The Problem, According to DatabricksIt is that companyies are still relaying on manual evaluations with a procese that’s account, incons are and different to scale.

Today at the data + AI Summit, Databricks Launcher Mosaic Agent Bricks As a Solution to that Challenge. The technology Builds on and ends the mosaic ai Agent Framework, Whiche the Company Annieded in 2024. Simply Put, it’s no longer Good En To Have a Real-World Impact.

The Mostaic Agent Bricks Platform Automates Agent Optimization Using a Series of Research-Backed Innovations. Among the Key Innovations is the Integration of Tao (Test-Time Adaptive Optimization), which provides a novel approach to ai tuning with the need for Labold data. Mosaic Agent Bricks always generates domain-simcific synthetic Data, Creates Task-AWARE BENCHMARKS and Optimizes Quality-To-Cost Balance With Manual Intervention.

The New Platform’s Fundamental Goal is to Solve an Isue that DATABRICKS USERS HAD With Existing Ai Agent Development Efforts.

“They Were Flying Blind, they had no way to evaluate these agents,“ Hanlin Tang, DATABRICKS ‘are “Most of them was relaying on A Kind of Manual, Manual VIBE Tracking to See if the Agent Sounds Good En

From Research Innovation to Enterprise Ai Production Scale

Tang Was Prevringly the Co-Founder and Cto of Mosaic, Who was Acquire by Databrics in 2023 for $ 1.3 Billion.

At Mosaic, Much of the Research Innovation DIDNNT NECESSARILY Have An An Imediate EnterPRISE IMPACT. That all changed after the acuision.

“The Big Light Bulb Momnt for Me Wheen We First Launched Our Product on Databrics, and Instantly, Overnight, We had, Like Thousands of Enterprise Customers Using It,” TANG said.

In card Integating Mosaic INTO DATABRICKS HAS Given Mosaic’s Research Team Direct Access to Enterprise Problems at Scale and Reveled New Areas to Explore.

This Enterprise Count Reveled New Research Opportunities.

“It’s only when you have received with enverrise Customers, you work with them Deeply, that you actially uncover of intelligesting research probems to go after,” Tang Explained. “Agent bricks….

Solving the Agentic Ai Evallation Crisis

Enterprise Teams Face A Costly Trial-End-Error Optimization Process. With task-awery benchmarks or domain-simcific Test data, every agent adjustment becomes An Expective Guessing Game. Quality Drift, Cost Overruns and Missed Deadlines Follow.

Agent Bricks Automats The Entire Optimization Pipeline. The Platform Takes A High-Level Task Description and Enterprise Data. IT Handles the Rest Automatically.

First, it generals task-simcific evalctions and llm judges. Next, It Creates Synthetic Data that Mirror Customer Data. Finally, It Searchches Across Optimization Techniques to find the Best Configuration.

“The Customer Descrabes The Problem at a High Level, and they don’t Go into “The System Generates Synthetic Data and Builds Custom Llm Judges Specific to Each Task.”

The Platform Offers Four Agent Configurations:

• Information Extraction: Converts Documents (pdfs, emails) INTO StruckRED Data. One Use Case Couelf Be Retail Organizations that Use it to Pull Product details from Supplier PDFS, Even with Complex Formatting.
• Knowledge Assistant: Provides Accurate, CITED ANSWERS from Enterprise Data. For Example, Manufacturing Technicians Can Get Instant Answeers from Maintenance Manuals with Digging Through Binders.
• CUSTOM LLMHandles Text Transformation Tasks (Summarization, Classification). For Example, Healthcare Organizations Can Customize Models that Summarize Patient Notes for Clinical Workflows.
• Multi-Agent SupervisorOrchestrates Multiple Agents for Complex Workflows. One Use Case Example is Financial Services FIRMS That Can Cooorate Agents for Intent Detection, Document Retrief and Compliance Checks.

Agents are Great, but Don’t Forget About Data

Building and Evaluating Agents Is a Core Part of Making An Ai Enterprise-Rady, but it’s not the Only Part That’s Needed.

Databricks Positions Mosaic Agent Bricks As the Ai Consump Layer Sitting Atop ITS Unified Data Stack. At the data + AI Summit, Databricks also Announceed the General Availability of Its Lakeflow Data Engineering Platform, Whiche was first Preview in 2024.

Lakeflow Solutions The Data Prepaction Challenge. It Unifies Three Critical Data Engineering Journeys that Prevringly Required Separate Tools. Ingestion handles getting Both Structures and UNSTRURURED DATA INTO DATABRICKS. Transformation Provides Efficience Data Cleang, Reshaping and Preparation. Orchestration Manages Production Workflows and Scheduing.

The Workflow Connection Is Direct: Lakeflow Prepares Enver Pata Through Unified Oncestion and Transformation, then Agent Bricks Builds Optimized Ai Agents on that Prepared Data.

“We help get the data into the playform, and then you can do ml, bi and ai Analytics,” Bilal Alam, Senior Director of Product Management at Databricks Told Venturebeat.

Going Beyond Data Oncestion, Mosaic Agent Bricks also benefits from Databricks’ Unity Catalog’s Goovernance Features. That Includs Access Controls and Data Lineage Tracking. This Integration Ensembles that Agent Beautiful Respects Enverrise Data Goovernance with Additional Configuration.

Agent Learning from Human Feedback Eliminates Prompt Stuffing

One of the Common Approaches to Guiding Ai Agents to Dyster is to use a system prombt. Tang Referred to the Practice of ‘Prompt Stuffing’ Where Users Shove All Kinds of Guidance Into A Prompt in the Hop that the Agent Will Follow it.

Agent Bricks Intduces A New Concept – Agent Learning from Human Feedback. This Feature Automatically Adjusts System Components Based on Natural Language Guidance. IT Solutions What Tang Calls The Prost Stuffing Problem. According to Tang, The Prost stuffing Approach often Fails Because Agent Systems Have Multiple Components that Need Adjustment.

Agent Learning from Human Feedback is a System that automatically interprets natural language guidance and adjusts The Appropriet System Components. The Approach Mirror Reinforce

The System Handles Two Core Challenges. First, Natural Language Guidance Can Be Vague. For Example, What Does ‘Respect Your Brand’s Voice’ Actually Mean? Second, Agent Systems Contain Numerous Configuration Points. Teams Struggle to Identify which components Need Adjustment.

The System Elimians The Guesswork About Which Agent Components Need Adjustment for Specific Behavioral Changes.

“This We Believe Will Help Agents Become More Sterable,” Tang SAID.

Technical Advantages Over Existing Frameworks

There is no Shortage of Agentic Ai Development Frameworks and Tools in the Market Toys. Among

Tang Argued that What Makes Mosaic Agent Bricks Difference is the Optimization. Ratter Than Requiring Manual Configuration and Tuning, Agent Bricks Incorporates Multiple Research Techniques Automatically: Tao, In-CONTEXT Learning, Prompt Optimization and Fine-Tuning.

There are a few Options for Agent-To-Agent Communications on the Market today, Including Google’s Agent2agent Protocol. According to Tang, Databricks is Currently Exploring Various Agent Protocols and Hasn’t Committed to A Single Standard.

Currently, Agent Bricks Handles Agent-To-Agent Commoniable Through Two Primary Methods:

1. Exposition Agents as Endpoints that can be Wrapped in different protocols.
2. Using A Multi-Agent Supervisor that Is McP (Model Contentxt Protocol) AWARE.

Strategic Implications for Enterprise Decision-Makers

For Entembriss Looking to Lead the Way in Ai, It’s Critical to Have the Right Technologies in Place to Evaluate Quality and Effective.

DePloying Agents With Evallation Will Not LEAD to An Optimal Outcome, and Neter Will Having Agents with a Solid Data Foundation. Where Considering Agent Development Technologies, It’s Critical to Have Proper Mechanisms to Evaluate the Best Options.

The Agent Learning from Human Feedback Approach Iso Noteworthy for Enterprise Decision Makers as it helps to Guide Agentic Ai to the Best outcome.

This Development Means Evallation Infrastructure is no longer a blocking factor for entertainris Looking to lead in Ai Agent DePloyment. Organizations Can Focus Resources on Use Case IDENTIOTION and Data Preparation Ratter Than Building Optimization Frameworks.