Databricks Agent Bricks
Agents on your Databricks data, built and deployed with Agent Bricks.
Agent Bricks is Databricks' platform for building agents in code and running them in production next to your governed data. Birbal AI builds the agent in the framework that fits, connects models, tools and memory, measures its quality with MLflow and deploys it on Databricks, with Genie Agents for the questions that need no code.
- Engineers based in the UAE
- In your Databricks workspace
- Fixed-price or team-based
Agent Bricks
DatabricksAuthorized vendor to
- Ministry of Finance
- Ministry of Interior
- DGE Abu Dhabi
- Etihad Aviation
- EDGE Group
Delivered for
- Hindustan Unilever
- Nokia
- Linde
- USA Today
- JSL
Procurement-ready
- Emirati national company
- National ICV certificate
- Abu Dhabi Government Procurement Gate
- UAE Ministry of Finance, federal procurement
- Registered ADNOC vendor
What we set up
Agents in code, run and governed next to your data
Use case and data
The agent's job, the Unity Catalog data it may use and how its answers will be measured.
Build in code
The agent scaffolded with the Agent Bricks CLI from a LangGraph or OpenAI Agents SDK template, or in your own framework.
Models and tools
Models called through Unity Gateway, so they can be switched without code changes, and tools over MCP and Unity Catalog functions.
Memory and sessions
Managed memory and session stores, so the agent keeps context across conversations.
Quality
MLflow tracing of every step, with LLM judges, custom scorers and your experts' feedback before go-live.
Deploy and govern
Deployment to Databricks' agent runtime, with access, guardrails, rate limits and usage tracked in Unity Gateway.
What it costs
What drives the cost of Agent Bricks
From Databricks' documentation and pricing page, checked on 20 November 2026. Rates differ by cloud and tier, so we estimate them with you.
Agent runtime
Agents run on Databricks-hosted serverless compute through Databricks Apps, on the number of instances you set when you deploy.
Model calls
Agents call Databricks-served models, and frontier or open models, through Unity Gateway, where usage is tracked and rate limits are set.
Beta and legacy
Agent Bricks is in Beta and needs no workspace setting; Databricks' earlier Knowledge Assistant and Supervisor Agent are not available to new customers.
Rates
Databricks lists Agent Bricks rates on its pricing page; its Premium tier on Azure corresponds to the Enterprise tier on AWS and Google Cloud.
How we work
From a shortlist to a governed rollout
- 01
Discovery
Your Databricks workspace and cloud, the data in Unity Catalog and the agent's job.
- 02
First agent
One agent for one team, built and tested on real cases before anyone else sees it.
- 03
Connect
Models, MCP tools and memory connected, with tracing and evaluations in MLflow.
- 04
Govern and launch
Permissions, logging and review, then a wider launch with training.
- 05
Run and improve
Monthly reviews of answers, usage and cost, then the next agents on the list.
Why Birbal AI
Why teams bring us in for Databricks
Engineers based in the UAE
An Emirati national company in Abu Dhabi, with engineers who work onsite in the UAE or remotely for teams anywhere.
Your tenant, your data
Agents are built in your own accounts and cloud, under your data rules.
People stay in charge
Agents hand over to your staff when they are unsure, and every action they take is logged.
You own the work
Configurations, code and documentation are transferred to you on delivery, with NDAs as standard.
AI agents in your software
Agents on the platform you already run
Sources
Where the facts on this page come from
The vendors' own pages, checked on 20 November 2026. Vendors change features, names and prices often; we check them again with you before you buy.
Free estimate
Get a free Databricks Agent Bricks estimate in 48 hours
Tell us which agents you need and the systems they should use. You get a scope, timeline and cost range within 48 hours, then a fixed-price proposal after a short discovery.
- Free and no obligation
- Fixed-price or team-based options
- NDA on request
4.9/5 average rating · 98% client satisfaction · 150+ clients served
FAQ
Databricks Agent Bricks: questions teams ask
What is Databricks Agent Bricks?
Databricks' agent developer platform: the infrastructure to build agents in code and run them in production, with model and tool access, deployment, memory, tracing, evaluation and governance on the same platform as your data.
How is Agent Bricks different from Genie Agents?
Genie Agents (formerly Genie spaces) are a low-code way to build agents that answer questions over your tables and documents. Agent Bricks is for writing and deploying your own agent code, and a custom agent can call Genie Agents as tools.
Which frameworks and models can we use?
Databricks says Agent Bricks works with any framework or harness. Its CLI includes templates for LangGraph and the OpenAI Agents SDK, and models are called through Unity Gateway, so you can switch them without changing agent code.
What happened to Knowledge Assistant and Supervisor Agent?
Databricks calls them legacy offerings and no longer offers them to new customers. For new low-code agents it points to Genie Agents, and for code-based agents to the Agent Bricks CLI.
Where do the agents run?
On Databricks' agent runtime, which runs your agent server on Databricks Apps serverless compute, with durable runs that can recover after a restart.
Do the agents work in Arabic?
Our team works in Arabic and English. Language quality depends on the model you choose, so we test Arabic with your own data and documents before rollout.
Agents on your Databricks data, in production
Get a scope, timeline and cost range within 48 hours.
Get a free estimate