Welcome to the Official Global Genie Community! 🧞✨

Databricks Genie change our vision for consuming and generate our informacion with natural language.

Help the business user like this "how is the best customer'.

And help the data steward "what's the Entidad Relacion Dimentional model of my schema table"

And help the data engineer like this 'What's is the linaje and the pipeline for load this field'

And the data priduct owner "how i make a fraud data product'

All this answer with Genie 💪

Databricks Genie is definitely a wonderful tool that helps the business users to get quick insights about the data.
I am confident in presenting the use case I have worked in the partner labs/community edition to the Business Users.

I am looking for guidance on how to present the cost that would be incurred for using the Genie by Business User.

Hey everyone! 👋

Currently migrating ~40 Power BI reports from a Dax based Jira cube to Databricks SQL on Unity Catalog.

Two things I'm here to explore:

1. Can Genie replace manual DAX-to-SQL conversion?

I've been writing every conversion by hand. There has to be a better way, Genie feels like the answer. Has anyone actually done this end to end? Would love to hear!

2. Chat on top of migrated reports

Once migration is done, business users should just ask questions in plain English. No more BI ticket queues.

Biggest question I have, how do you get Genie to respect business logic, not just write syntactically correct SQL?

Excited to learn from this community! 🙌

I've been impressed with Genie's capabilities for few months now.
Probably like most people I mostly played around with: Natural language built-in support to getting insights, building quick dashboards and writing notebooks code for ETL pipelines.

The harder question is how to turn that into a trusted, enterprise-grade solution at scale. A few worth to discuss at th beginning:

• Where does Genie (not only Ask mode) genuinely shine, and where is it overkill?

• How do you avoid the "self-service analytics everywhere" trap that burns through DBUs without delivering proportional value?

• What does a governed, production-ready Genie setup actually should look like?

Looking forward to real-life implementation stories, especially from people who've hit these walls.

P.S. I used Genie to help draft this post, which felt like a fitting way to start (human)

Hello folks!!

my challenge with Ginie now is to create MCPs for each tool in our ecosystem tools, like datafactory, azure devops, etc

Hi Everyone,

Great to be a part of this community! I'm looking forward to all the interesting use cases posted here and sharing a few that I am planning to work on.

Pradeep Singh
https://www.linkedin.com/in/dbxdev/

I would be interested to know how Genie will assist in generating useful insights on telco statistical data enabling network optimizers to gain a different perspective of the end user QoS levels and along with current traffic pattern and possible loopholes in network.

One of the most impressive capabilities in the Databricks ecosystem is Databricks Genie. It transforms the way users interact with data by enabling natural language conversations directly with enterprise datasets.

Instead of writing complex SQL queries or navigating multiple dashboards, users can simply ask questions in plain English and receive accurate, contextual insights within seconds. This significantly reduces the barrier to data access and empowers business users, analysts, and decision-makers to become more data-driven.

What stands out is how Genie combines the power of AI with the governance, security, and scalability of the Databricks Data Intelligence Platform. It's a great example of how generative AI can make analytics more accessible, efficient, and impactful across organizations.

A big step toward truly democratizing data!
#Databricks #DatabricksGenie #DataIntelligence #GenerativeAI #Analytics #DataDriven

🚀 From Curiosity to Impact: My Databricks Genie Experience

I’ve been a Databricks Champion since 2025, and I still remember the early days when I presented a POC to the panel—knowing just the basics of what Genie and Genie Space could do.

Fast forward to today, and I finally got the chance to explore its true potential within a real enterprise use case.

Recently, one of our product teams built an end-to-end Databricks Dashboard:

• 📊 ~11 pages

• 📈 ~80 KPIs

• ⚙️ Complex backend queries

My task?
To deep dive into each page and identify opportunities to standardize and optimize it for centralized observability.

Instead of going through hours of manual analysis, I tried something different.

💡 The Experiment

I prompted Genie Code with:

“Go through the entire dashboard, build a markdown document explaining each KPI per page — include 2 lines on what each KPI does, table dependencies, columns used, duplicates, and possible enhancements.”

⚡ The Outcome

In just 8 minutes (on XXS SQL Compute), Genie generated a comprehensive, structured document covering everything I needed.

✅ Clear explanation of all KPIs
✅ Dependencies and column-level insights
✅ Identification of duplication opportunities
✅ Practical enhancement suggestions

And most importantly…

⏱️ It saved me nearly 6 hours of manual effort.

🤯 Why This Matters

At first glance, this may seem like a small win. But if you've worked with tools like:

• Power BI

• Qlik

• Tableau

…you know that this kind of deep documentation and analysis is almost entirely manual.

What took hours of human effort can now be done in minutes with AI assistance—and with impressive accuracy.

🔍 The Bigger Picture

This isn’t just about saving time.

It’s about:

• Enabling faster insights

• Driving standardization at scale

• Reducing human error

• Unlocking true data productivity

We’re moving from dashboard consumption to intelligent data understanding.

🔥 Final Thought

Sometimes we underestimate use cases like this—until we experience their impact firsthand.

Genie isn’t just a feature.
It’s a force multiplier.

After pushing Genie hard for the last 6 months, I can honestly say it’s a great product.

Before trying Genie Spaces, I built my own SQL agent. It worked, but not nearly as well as what Databricks has built. What stood out most through all of this testing was not just how capable Genie is, but what it reveals.

Good AI products do not fix bad data. They expose it.

Our biggest challenge was never the agent itself. It was getting the foundation truly AI-ready: clean schemas, reliable pipelines, clear business definitions, and data that is actually easy to query.

That’s why I’ve become such a big believer in metric views. When the metric view is well documented and captures the business nuance correctly, Genie is 98% of the way there. At that point, it can give reliable answers with a level of consistency that is hard to get otherwise.

Then the last mile becomes much easier. If I want Genie to change tone, ask follow-ups, or explain what it’s doing, that’s where text instructions really shine.

If I had one ask to become a true Genie Genius, it would be a little more transparency. It would be incredibly helpful to better understand how Genie itself is instructed so my instructions don’t conflict with the system, which is one of the fastest ways to confuse an LLM.

My second ask would be more examples. Databricks has great sample data and sandbox environments, but I would love to see a sample Genie Space that is truly mind-blowing.

exciting stuff! I'm very happy that I can connect to external MCPs! 🚀

keen on getting MCP Apps working inside Genie too 👀!

Excited to be part of the #GenieCommunity!

As someone passionate about AI and analytics, I'm excited to learn how teams are making Genie a trusted part of everyday decision-making.

I'd love to explore use cases around customer care center, operations, and product analytics where business users can get answers without waiting on reports.

Curious to hear what prompts, lessons learned, and adoption strategies have worked best for others.

Looking forward to becoming a #GenieGenius with this community!

Sharing a great customer story with Genie that we have run over the past few months.

Really excited to see adoption picking up, from different places of the organization, in a distributed way that really enables the end-users.

https://www.linkedin.com/pulse/bridging-self-service-bi-reporting-genie-spaces-kristian-johannesen-osfte/

Happy to be part of this community! Genie has been a great addition into our org's lakehouse. One of the most interesting part which has always made me curious while working with Databricks genie is how would I get to know how much does these chats cost actually. We have those cost tracking present with other CSP's AI models which ofcourse we use indirectly via Keys or MIs. But is there a way to know how much I am burning with Databricks genie so that we could compare the benefits quantifiably?

Hi Folks
Sheetesh this side

Let's grow this Databricks Community together..

As a complete beginner in Databricks, I am curious about how Genie provides competitive adventage and how it benefits professionals in data field. 🍀

Thanks for Adding us.

Genie is Great tool when it comes to fixing Bugs, Building prototype, Refactoring solutions in minutes.

Happy to be part of this community.

Over the past few months I’ve been pushing Genie far beyond simple NL-to-SQL use cases, and it’s been eye‑opening to see how much value it can unlock when paired with clean semantic layers and well‑defined business logic. One of the biggest breakthroughs for me came when I started treating Genie not just as a query assistant, but as an intelligence layer on top of our entire data estate.

For example, I’ve used Genie to validate metric definitions, trace lineage across complex pipelines, and even stress‑test our dimensional models by prompting it with edge‑case business questions — something others in this thread have also highlighted as a real challenge when scaling Genie to enterprise‑grade use cases . What surprised me most is how quickly Genie exposes inconsistencies in schemas and definitions, echoing what Jack mentioned about AI revealing data quality gaps rather than masking.

One of my heavier experiments involved feeding Genie structured instructions to generate end‑to‑end documentation for multi‑page dashboards — KPI logic, dependencies, duplication patterns, and enhancement opportunities. Similar to Somesh’s experience, this turned hours of manual analysis into minutes of automated insight, and the consistency of the output was impressive .

My biggest ongoing focus is making Genie trustworthy at scale: ensuring it respects business logic, aligns with governed metric views, and produces repeatable results even as datasets evolve. When those foundations are solid, Genie becomes a genuine force multiplier for analysts, engineers, and business teams alike.

Excited to learn from others here and continue pushing the boundaries of what Genie can do.

We are currently in the process of productionizing 3 Genie spaces for our users and believe me it has been a WONDERFUL journey so far working on Genie spaces and metric views.
Its such an intuitive tool and will really help a lot of non-tech users to get their KPIs on fingertips. Its great how you can learn so much by just working on something. Everyday we add something new to the genie spaces in terms of guard rails or working instructions.

I have also built a CI/CD reusable framework to migrate genie spaces either 1. workspace to workspace OR 2. catalog to catalog.

If you want to check it out please visit the links below:
1. Workspace to Workspace : https://github.com/yashojha1/Databricks-Genie-Migration

2. Catalog to Catalog : https://github.com/yashojha1/genie-catalog-migration

Happy to connect, learn and grow together !!! Cheers

Hello Everyone, Happy to part of this community

What’s the most interesting Genie use case you’ve built (or want to build)?
I began using the Genie Public Preview in October while working on an enterprise logistics migration. Initially I used Genie for general error debugging. After seeing many use cases on LinkedIn and Medium—coding, documentation, and analytic Q&A—I started using pre-built frameworks and conversion code from third-party AI, importing the results into Databricks and using Genie to debug and refine them.

We ran POCs to migrate complex stored procedures, ADF pipelines, and dataflows to Genie using curated prompts to convert code into Databricks notebooks with relevant functions and logic. After two or three iterations, Genie produced well-structured documentation (Markdown files, cell descriptions, dynamic variables). We validated the code by comparing results and updated the Genie-generated markdown for edge cases. By late November Genie’s reasoning and decision-making had improved significantly. Over the last three months we relied on Genie to complete the migration on schedule with efficient code.

Why Genie worked better than third-party APIs for us
Genie has full context of notebooks, catalogs, jobs, and dashboards, so it can produce more accurate outputs than other AI tools that lack that integrated context.

Recent win: report migration
We recently needed to migrate reports to Databricks. Using the AI Dev Kit, we built custom skills and instructions to feed Genie Power BI exports (.tts, .pbix) plus page screenshots. Genie rebuilt about 95% of each PBI report into an AI/BI dashboard; we only adjusted alignments and some metric logic. This was a major win for the reporting team.

What’s your biggest challenge with Genie today?
Genie updates rapidly. It would help to have dedicated documentation pages or a GitHub repo detailing steps and options so users can fully leverage its capabilities.

What tips, tricks, prompts, or best practices should the community share?
There are many techniques to get near-perfect outputs from Genie: refining instructions, breaking work into task-based items, crafting prompts that limit unnecessary context, and tailoring prompts for specific migration edge cases. For our logistics use case we documented 30+ edge cases covering MSSQL, PostgreSQL, and MySQL migrations.

What would help you become a true Genie expert?
Staying up to date with Databricks releases, exploring more use cases across verticals, and following community updates on LinkedIn.

Thanks, Samantha for starting the Genie Community. I look forward to sharing tips and learning from the community.

Hi Genie community, Genie is developing rapidly, and if you would like to receive the latest updates on Genie every month, feel free to subscribe to our Substack: https://aibilakehouse.substack.com/ :)

I’ve been exploring Databricks Genie recently, and it genuinely feels like a shift from “querying data” to “conversing with data.”

What stands out to me is how Genie bridges multiple personas in the same platform:

• Business users can ask natural language questions and get instant insights

• Data engineers can understand lineage, pipelines, and transformations faster

• Data stewards get better visibility into schema relationships and definitions

But like many have mentioned, Genie doesn’t replace strong data foundations — it amplifies them. Clean data models, governed metrics, and clear semantic layers are what make Genie truly powerful.

In my experience, the real value comes when Genie is layered on top of well-defined data products and metric views — that’s when responses become consistent, reliable, and actionable.

I also see huge potential in integrating Genie across the broader ecosystem (ADF, DevOps, etc.), especially to create end-to-end intelligence from ingestion to consumption.

Overall, Genie is not just a feature — it’s a step toward making data platforms more intuitive and accessible at scale 🚀