Databricks for Startups (@DBXforStartups) header
Databricks for Startups (@DBXforStartups) profile picture

Databricks for Startups offers credits, expert advice, and market reach for VC-funded, data-driven startups.

San Francisco, CA
Joined September 2025
Photo by Databricks for Startups (@DBXforStartups)Photo by Databricks for Startups (@DBXforStartups)Photo by Databricks for Startups (@DBXforStartups)Photo by Databricks for Startups (@DBXforStartups)Photo by Databricks for Startups (@DBXforStartups)Photo by Databricks for Startups (@DBXforStartups)Photo by Databricks for Startups (@DBXforStartups)Photo by Databricks for Startups (@DBXforStartups)
Join the Databricks and Mayfield team for a dinner on Wednesday, March 18 in San Jose โ€“ a night designed exclusively for startup leaders and executives who are shaping the future of data & AI. ๐Ÿš€ โ€‹โ€‹Weโ€™re bringing together the boldest minds in the AI startup ecosystem for an intimate dinner where youโ€™ll have the chance to network, share best practices, and learn how others are using Databricks. Seats are limited. Donโ€™t miss this opportunity to join one of the most vibrant founder communities around GTC! ๐Ÿ‘‰ Secure your spot at the table here: events.databricks.com/FY27Q1โ€ฆ #Databricks #Founders #Startups #DatabricksForStartups
1
83
Calling all LA food-tech builders & operators! ๐ŸŒด Weโ€™re excited to partner with Topi for "Topi Topics Live." Itโ€™s a day for restaurant operators, founders, and the AI-curious to get into the weeds of how tech is changing culture. Whatโ€™s happening: ๐Ÿ’ป ๐—ฉ๐—ถ๐—ฏ๐—ฒ ๐—–๐—ผ๐—ฑ๐—ถ๐—ป๐—ด: Prototype tools live with Databricksโ€”no coding skills needed. โšก ๐—Ÿ๐—ถ๐—ด๐—ต๐˜๐—ป๐—ถ๐—ป๐—ด ๐—ง๐—ฎ๐—น๐—ธ๐˜€: Fast-paced, high-signal PechaKucha-style insights (20 slides ร— 20 seconds). ๐Ÿค ๐—–๐—ผ๐—ป๐—ป๐—ฒ๐—ฐ๐˜๐—ถ๐—ผ๐—ป๐˜€: Real talk and real time to build community. ๐Ÿ—“๏ธ Monday, March 2nd ๐Ÿ“ Los Angeles, CA ๐Ÿ”— RSVP: luma.com/uhacglc9?tk=hqF9Tt Food and drinks will be served throughout, but space is limited. Hope to see you there! ๐Ÿ‘‹ #Databricks #FoodTech #Hospitality #Startups #DatabricksForStartups
46
๐Ÿ“ฃ Join the @databricks VC-Backed Startups team and SignalFire on Thursday, March 12 in downtown NYC! Weโ€™re hosting an intimate dinner designed specifically for the founders and execs who are actually shaping the future of data & AI. ๐Ÿš€ Network, share best practices, and learn what others are who are using data to build whatโ€™s next. โ€‹โ€‹Seats are limited โ€“ and this community moves fast. Donโ€™t miss this opportunity to join one of the most vibrant founder communities in NYC. ๐Ÿ™๏ธ Register here ๐Ÿ‘‰ luma.com/nyc-startupdinner31โ€ฆ #Databricks #SignalFire #NYCStartups #DatabricksForStartups
1
2
100
Weโ€™re proud to be a Founding Host of Bostonโ€™s first-ever Tech Week! ๐ŸŽ‰ Databricks for Startups is joining 75+ innovative companies to bring a massive week of programming to Boston from ๐Ÿ—“๏ธ May 26 โ€“ May 31. Want to host an event alongside top companies and leaders shaping the future of tech? Event submissions are open now at ๐Ÿ”— tech-week.com/host #Databricks #DatabricksForStartups #BOSTechWeek @Techweek_
32
We were thrilled to be part of the 4th Global AI Hackathon, the world's premier AI innovation challenge! ๐Ÿ™Œ It was a privilege to watch brilliant minds converge to build the future of AI in just one weekend. Judging at Stanford and MIT, our team was blown away by the creativity, the technical depth of the projects, and the incredible geographical diversity of the participants. ๐ŸŒ Congratulations to every team that participated! The speed and innovation we saw in just 24 hours prove that the next generation of AI startups is already building the future. ๐Ÿš€ #Databricks #Stanford #MIT #GlobalAIHackathon #AIInnovation #Startups #DatabricksForStartups #Hackathon
43
Join the @databricks VC-Backed Startups team and @jpmorgan for an intimate dinner on ๐—ง๐—ต๐˜‚๐—ฟ๐˜€๐—ฑ๐—ฎ๐˜†, ๐—™๐—ฒ๐—ฏ๐—ฟ๐˜‚๐—ฎ๐—ฟ๐˜† ๐Ÿญ๐Ÿต, in central Manhattan, NYC. ๐Ÿ™๏ธ Weโ€™re bringing together the boldest minds in NYCโ€™s startup ecosystem. ๐Ÿš€ Expect honest conversation, shared war stories, and perspectives from an array of industries and backgrounds. Who's this for? โ€‹๐Ÿ”น ๐—ฉ๐—ถ๐˜€๐—ถ๐—ผ๐—ป๐—ฎ๐—ฟ๐—ถ๐—ฒ๐˜€: Early-stage startup founders at the cusp of innovation โ€‹๐Ÿ”น ๐—•๐˜‚๐—ถ๐—น๐—ฑ๐—ฒ๐—ฟ๐˜€: Technical leaders in fast-scaling companies ๐Ÿ”น โ€‹๐—ฆ๐—ฒ๐—ฎ๐˜€๐—ผ๐—ป๐—ฒ๐—ฑ ๐—ฃ๐—ฟ๐—ผ๐—ณ๐—ฒ๐˜€๐˜€๐—ถ๐—ผ๐—ป๐—ฎ๐—น๐˜€: Entrepreneurs passionate and involved in NYC's data, AI, and open source tech scene โ€‹โ€‹Seats are limited โ€“ and this community moves fast. Donโ€™t miss this opportunity to join one of the most vibrant founder communities in NYC. ๐Ÿ‘‡ ๐Ÿ”— Register: luma.com/nyc-startupdinner21โ€ฆ #databricksforstartups #startups #databricks #nyc
41
๐Ÿšจ New Blog: ๐—›๐—ถ๐—ฑ๐—ฑ๐—ฒ๐—ป ๐—ง๐—ฒ๐—ฐ๐—ต๐—ป๐—ถ๐—ฐ๐—ฎ๐—น ๐——๐—ฒ๐—ฏ๐˜ ๐—ผ๐—ณ ๐—š๐—ฒ๐—ป๐—”๐—œ ๐—ฆ๐˜†๐˜€๐˜๐—ฒ๐—บ๐˜€ Classical ML and GenAI are different flavors of the same domain. While you may be used to spending weeks on data cleaning, GenAI flips the script, shifting effort toward subjective evaluation and stakeholder management. In this blog, youโ€™ll learn: ๐Ÿ”„ย ๐——๐—ฒ๐˜ƒ๐—ฒ๐—น๐—ผ๐—ฝ๐—บ๐—ฒ๐—ป๐˜ ๐—™๐—น๐—ถ๐—ฝ: Why GenAI development time is dominated by evaluation and gathering human-in-the-loop feedback โš ๏ธย  ๐Ÿฑ ๐—›๐—ถ๐—ฑ๐—ฑ๐—ฒ๐—ป ๐——๐—ฒ๐—ฏ๐˜๐˜€: Dive deep into Tool Sprawl, Prompt Stuffing, and Opaque Pipelinesโ€”plus the risks of inadequate feedback loops and stakeholder silos โœ…ย ๐—ช๐—ถ๐—ป๐—ป๐—ถ๐—ป๐—ด ๐—ฆ๐˜๐—ฟ๐—ฎ๐˜๐—ฒ๐—ด๐—ถ๐—ฒ๐˜€: Why clear communication and robust monitoring remain the only ways to build clean, maintainable systems ๐Ÿ”— Dive in: databricks.com/blog/hidden-tโ€ฆ #Databricks #Startups #GenAI #LLMOps
18
As AI-native products transition from pilots to mission-critical workflows, the competitive advantage is shifting towards data and AI workflow architecture. Join us for an evening of networking and discussion, co-hosted with G2C Ventures, for GenAI founders and senior product, data, and engineering leaders to go deeper on how modern data lakes enable scalable, cost-efficient, and defensible Data + AI strategies. ๐Ÿ’ก ๐Ÿ—“๏ธ When:ย Thursday, January 15th at 5:00 PM ๐Ÿ‘ฅ Who:ย GenAI Native founders, Directors/VPs at <$200M ARR ISVs ๐Ÿ“ Where:ย Palo Alto, CA Why attend? ๐Ÿ”น Learn how leading teams are structuring their data foundations for GenAI= ๐Ÿ”น Pressure-test your Data + AI roadmap with peers at a similar stage ๐Ÿ”น Exchange ideas on whatโ€™s working and whatโ€™s not ๐Ÿ”น Build relationships with founders and executives solving adjacent problems ๐Ÿ”— REGISTER: luma.com/kcmmmf7j #genai #databricks #databricksforstartups #datalakes
24
๐——๐—ฎ๐˜๐—ฎ๐—ฏ๐—ฟ๐—ถ๐—ฐ๐—ธ๐˜€ ๐—ณ๐—ผ๐—ฟ ๐—ฆ๐˜๐—ฎ๐—ฟ๐˜๐˜‚๐—ฝ๐˜€ ๐—ค๐Ÿฐ, ๐Ÿฎ๐Ÿฌ๐Ÿฎ๐Ÿฑ ๐—ฅ๐—ฒ๐˜ƒ๐—ถ๐—ฒ๐˜„ ๐Ÿš€ Our latest quarterly recap covers the launch of the Databricks AI Accelerator, our founder dinner series, and more. See what weโ€™ve been up to! โžก๏ธ linkedin.com/pulse/databrickโ€ฆ #databricks #databricksforstartups #founders #startups
56
๐—ฆ๐—ฎ๐—ป๐˜๐—ฎ ๐—ฑ๐—ผ๐—ฒ๐˜€๐—ป'๐˜ ๐—ฐ๐—ต๐—ฒ๐—ฐ๐—ธ ๐˜๐—ต๐—ฒ ๐—น๐—ถ๐˜€๐˜ ๐˜๐˜„๐—ถ๐—ฐ๐—ฒ ๐—ฎ๐—ป๐˜†๐—บ๐—ผ๐—ฟ๐—ฒ. ๐—›๐—ฒ ๐—ฐ๐—ต๐—ฒ๐—ฐ๐—ธ๐˜€ ๐—ฎ ๐—ฑ๐—ฎ๐˜€๐—ต๐—ฏ๐—ผ๐—ฎ๐—ฟ๐—ฑ. ๐ŸŽ… We have reached the finish line! Itโ€™sย ๐——๐—ฎ๐˜† ๐Ÿญ๐Ÿฎ of the ๐Ÿญ๐Ÿฎ ๐——๐—ฎ๐˜†๐˜€ ๐—ผ๐—ณ ๐——๐—ฒ๐—บ๐—ผ๐˜€, and we are replacing manual checks with real-time visibility at the North Pole. Over the last 11 days, we ingested streaming data, built vector indexes, deployed AI agents, managed Postgres instances, and more. But the North Pole leadership team doesn't look at raw tables or Python notebooks. They look atย ๐——๐—ฎ๐˜€๐—ต๐—ฏ๐—ผ๐—ฎ๐—ฟ๐—ฑ๐˜€. Usually, this is where you move dataย outย of your platform into a third-party BI tool, rebuilding logic with complex semantic models and losing governance along the way. ๐—ง๐—ผ๐—ฑ๐—ฎ๐˜†, ๐˜„๐—ฒ ๐˜ƒ๐—ถ๐˜€๐˜‚๐—ฎ๐—น๐—ถ๐˜‡๐—ฒ๐—ฑ ๐—ฒ๐˜ƒ๐—ฒ๐—ฟ๐˜†๐˜๐—ต๐—ถ๐—ป๐—ด ๐—ฟ๐—ถ๐—ด๐—ต๐˜ ๐˜„๐—ต๐—ฒ๐—ฟ๐—ฒ ๐—ถ๐˜ ๐—น๐—ถ๐˜ƒ๐—ฒ๐˜€. Using ๐——๐—ฎ๐˜๐—ฎ๐—ฏ๐—ฟ๐—ถ๐—ฐ๐—ธ๐˜€ ๐—”๐—œ/๐—•๐—œ ๐——๐—ฎ๐˜€๐—ต๐—ฏ๐—ผ๐—ฎ๐—ฟ๐—ฑs, we build a view into operations for North Pole leadership. ๐—ช๐—ต๐˜† ๐˜๐—ต๐—ถ๐˜€ ๐—ถ๐˜€ ๐˜๐—ต๐—ฒ ๐—ฝ๐—ฒ๐—ฟ๐—ณ๐—ฒ๐—ฐ๐˜ ๐—ฒ๐—ป๐—ฑ ๐˜๐—ผ ๐˜๐—ต๐—ฒ ๐˜€๐˜๐—ฎ๐—ฐ๐—ธ: ๐Ÿ”น ๐—๐˜‚๐˜€๐˜ ๐—ฆ๐—ค๐—Ÿ:ย No complex semantic modeling layers. If you can query it, you can graph it. ๐Ÿ”น ๐—•๐˜‚๐—ถ๐—น๐˜-๐—ถ๐—ป ๐—”๐—œ:ย We used the AI_FORECAST function to predict toy production trends directly on the chart. ๐Ÿ”น ๐—š๐—ผ๐˜ƒ๐—ฒ๐—ฟ๐—ป๐—ฒ๐—ฑ:ย We didn't have to recreate security rules. The dashboard respects the exact same Unity Catalog permissions as the rest of the Lakehouse. ๐—ง๐—ต๐—ฒ ๐—Ÿ๐—ฎ๐—ธ๐—ฒ๐—ต๐—ผ๐˜‚๐˜€๐—ฒ ๐—ถ๐˜€ ๐—ฐ๐—ผ๐—บ๐—ฝ๐—น๐—ฒ๐˜๐—ฒ.ย From raw letters to AI agents to executive dashboards: one platform, one governance model, zero silos. Oh, and all of this is ready to consume through the newย ๐——๐—ฎ๐˜๐—ฎ๐—ฏ๐—ฟ๐—ถ๐—ฐ๐—ธ๐˜€ ๐—ข๐—ป๐—ฒย business-facing hub of the platform (more in the article). Thank you for following along with the North Pole Modernization Office! ๐Ÿ—“๏ธ Start Day 12:ย medium.com/startups-at-databโ€ฆ ๐Ÿ”” Missed previous days? Catch all the demo drops here:ย medium.com/startups-at-databโ€ฆ ๐Ÿ’ป Get the Code:ย github.com/arjuncode/12daysoโ€ฆ #Databricks #DataVisualization #BI #AI #Lakehouse #DataEngineering #DatabricksOne
14
๐—ช๐—ฒ ๐—ฏ๐˜‚๐—ถ๐—น๐˜ ๐—ฎ ๐—–๐—ผ๐—บ๐—บ๐—ฎ๐—ป๐—ฑ ๐—–๐—ฒ๐—ป๐˜๐—ฒ๐—ฟ ๐—ฎ๐—ฝ๐—ฝ ๐—ณ๐—ผ๐—ฟ ๐˜๐—ต๐—ฒ ๐—ก๐—ผ๐—ฟ๐˜๐—ต ๐—ฃ๐—ผ๐—น๐—ฒ ๐˜„๐—ถ๐˜๐—ต๐—ผ๐˜‚๐˜ ๐˜„๐—ฟ๐—ถ๐˜๐—ถ๐—ป๐—ด ๐—ฎ ๐˜€๐—ถ๐—ป๐—ด๐—น๐—ฒ ๐——๐—ผ๐—ฐ๐—ธ๐—ฒ๐—ฟ๐—ณ๐—ถ๐—น๐—ฒ ๐—ผ๐—ฟ ๐—บ๐—ฎ๐—ป๐—ฎ๐—ด๐—ถ๐—ป๐—ด ๐—ฎ๐—ป๐˜† ๐—ถ๐—ป๐—ณ๐—ฟ๐—ฎ๐˜€๐˜๐—ฟ๐˜‚๐—ฐ๐˜๐˜‚๐—ฟ๐—ฒ. Itโ€™sย ๐——๐—ฎ๐˜† ๐Ÿญ๐Ÿญย of theย ๐Ÿญ๐Ÿฎ ๐——๐—ฎ๐˜†๐˜€ ๐—ผ๐—ณ ๐——๐—ฒ๐—บ๐—ผ๐˜€. ๐ŸŽ„ Data pipelines are great, but the Elves can't use a SQL notebook to mark a package as "Delivered." They need buttons. They need maps. Usually, building this means spinning up external infrastructure and wrestling with authentication. ๐—ง๐—ผ๐—ฑ๐—ฎ๐˜†, ๐˜„๐—ฒ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐—ฎ๐—ป๐—ฑ ๐—ฑ๐—ฒ๐—ฝ๐—น๐—ผ๐˜† ๐˜๐—ต๐—ฒ "๐—ก๐—ผ๐—ฟ๐˜๐—ต ๐—ฃ๐—ผ๐—น๐—ฒ ๐—–๐—ผ๐—บ๐—บ๐—ฎ๐—ป๐—ฑ ๐—–๐—ฒ๐—ป๐˜๐—ฒ๐—ฟ" ๐—ฎ๐—ฝ๐—ฝ ๐—ฑ๐—ถ๐—ฟ๐—ฒ๐—ฐ๐˜๐—น๐˜† ๐—ถ๐—ป๐˜€๐—ถ๐—ฑ๐—ฒ ๐——๐—ฎ๐˜๐—ฎ๐—ฏ๐—ฟ๐—ถ๐—ฐ๐—ธ๐˜€. Usingย ๐——๐—ฎ๐˜๐—ฎ๐—ฏ๐—ฟ๐—ถ๐—ฐ๐—ธ๐˜€ ๐—”๐—ฝ๐—ฝ๐˜€, we took a standard Python (Dash) app script and deployed it as a secure, serverless web application. This is not a read-only dashboard. Because we connected it toย ๐—Ÿ๐—ฎ๐—ธ๐—ฒ๐—ฏ๐—ฎ๐˜€๐—ฒ (๐—ฃ๐—ผ๐˜€๐˜๐—ด๐—ฟ๐—ฒ๐˜€)ย from Day 10, the app is fullyย ๐—ฅ๐—ฒ๐—ฎ๐—ฑ/๐—ช๐—ฟ๐—ถ๐˜๐—ฒ ๐˜„๐—ถ๐˜๐—ต ๐—ฏ๐˜‚๐—ถ๐—น๐˜-๐—ถ๐—ป ๐—ด๐—ผ๐˜ƒ๐—ฒ๐—ฟ๐—ป๐—ฎ๐—ป๐—ฐ๐—ฒ. Zero Dockerfiles. Zero external hosting. Just App and Lakehouse working together. ๐Ÿ—“๏ธ Start Day 11:ย medium.com/startups-at-databโ€ฆ ๐Ÿ”” Missed previous days? Catch all the demo drops here:ย medium.com/startups-at-databโ€ฆ ๐Ÿ’ป Get the Code:ย github.com/arjuncode/12daysoโ€ฆ #Dash #Python #FullStack #DataEngineering #Databricks
1
1
43
๐—ง๐—ต๐—ฒ ๐—ก๐—ผ๐—ฟ๐˜๐—ต ๐—ฃ๐—ผ๐—น๐—ฒ ๐—ถ๐˜€ ๐—ฏ๐˜‚๐˜€๐˜† ๐—ฒ๐—ป๐—ผ๐˜‚๐—ด๐—ต ๐˜„๐—ถ๐˜๐—ต๐—ผ๐˜‚๐˜ ๐—ฑ๐—ฒ๐—ฏ๐˜‚๐—ด๐—ด๐—ถ๐—ป๐—ด ๐—ฐ๐—น๐˜‚๐—ป๐—ธ๐˜† ๐—ฅ๐—ฒ๐˜ƒ๐—ฒ๐—ฟ๐˜€๐—ฒ ๐—˜๐—ง๐—Ÿ ๐—ฝ๐—ถ๐—ฝ๐—ฒ๐—น๐—ถ๐—ป๐—ฒ๐˜€. Itโ€™sย ๐——๐—ฎ๐˜† ๐Ÿญ๐Ÿฌย of theย ๐Ÿญ๐Ÿฎ ๐——๐—ฎ๐˜†๐˜€ ๐—ผ๐—ณ ๐——๐—ฒ๐—บ๐—ผ๐˜€, and we have a classic integration problem at the North Pole. The Data Science team loves Spark. The Workshop Operations team needs a database connection. Usually, this means building a brittle "Reverse ETL" pipeline to shove data out of a secure platform into a siloed database. Today, we simplified the story with ๐—Ÿ๐—ฎ๐—ธ๐—ฒ๐—ฏ๐—ฎ๐˜€๐—ฒ, fully managed Postgres from @databricks. The real magic: Instead of writing complex sync code, we justย ๐—ฑ๐—ฒ๐—ฐ๐—น๐—ฎ๐—ฟ๐—ฒย a Synced Table (from a Delta table to Lakebase Postgres). Analytical power and operational speed, all in one platform. ๐Ÿ—“๏ธ Start Day 10:ย medium.com/startups-at-databโ€ฆ ๐Ÿ”” Missed previous days? Catch all the demo drops here:ย medium.com/startups-at-databโ€ฆ ๐Ÿ’ป Get the Code:ย github.com/arjuncode/12daysoโ€ฆ #Databricks #Lakebase #ETL #DataEngineering #FullStack
12
๐—ง๐—ต๐—ฒ ๐—ก๐—ผ๐—ฟ๐˜๐—ต ๐—ฃ๐—ผ๐—น๐—ฒ ๐—ถ๐˜€ ๐—ฎ ๐—ฐ๐—ผ๐—บ๐—ฝ๐—น๐—ฒ๐˜… ๐—ผ๐—ฟ๐—ด๐—ฎ๐—ป๐—ถ๐˜‡๐—ฎ๐˜๐—ถ๐—ผ๐—ป. ๐—ฌ๐—ผ๐˜‚ ๐—ฐ๐—ฎ๐—ป'๐˜ ๐—ฒ๐˜…๐—ฝ๐—ฒ๐—ฐ๐˜ ๐˜๐—ต๐—ฒ ๐— ๐—ฎ๐—ถ๐—น๐—ฟ๐—ผ๐—ผ๐—บ ๐—˜๐—น๐—ณ ๐˜๐—ผ ๐—ธ๐—ป๐—ผ๐˜„ ๐˜„๐—ต๐—ฎ๐˜'๐˜€ ๐—ต๐—ฎ๐—ฝ๐—ฝ๐—ฒ๐—ป๐—ถ๐—ป๐—ด ๐—ผ๐—ป ๐˜๐—ต๐—ฒ ๐—™๐—ฎ๐—ฐ๐˜๐—ผ๐—ฟ๐˜† ๐—™๐—น๐—ผ๐—ผ๐—ฟ. Itโ€™sย ๐——๐—ฎ๐˜† ๐Ÿต of the ๐Ÿญ๐Ÿฎ ๐——๐—ฎ๐˜†๐˜€ ๐—ผ๐—ณ ๐——๐—ฒ๐—บ๐—ผ๐˜€, and we are applying that same logic to our AI. A single AI agent can't be an expert at everything. If you ask one model to analyze theย sentimentย of a letter and thenย queryย a production database, it often gets confused. Itโ€™s too much context switching and optimization gets difficult. ๐—ง๐—ต๐—ฒ ๐˜€๐—ผ๐—น๐˜‚๐˜๐—ถ๐—ผ๐—ป? ๐—ช๐—ฒ ๐—ป๐—ฒ๐—ฒ๐—ฑ ๐—ฎ ๐—ฆ๐˜‚๐—ฝ๐—ฒ๐—ฟ๐˜ƒ๐—ถ๐˜€๐—ผ๐—ฟ. Today, usingย ๐——๐—ฎ๐˜๐—ฎ๐—ฏ๐—ฟ๐—ถ๐—ฐ๐—ธ๐˜€ ๐—”๐—ด๐—ฒ๐—ป๐˜ ๐—•๐—ฟ๐—ถ๐—ฐ๐—ธ๐˜€, we built aย ๐— ๐˜‚๐—น๐˜๐—ถ-๐—”๐—ด๐—ฒ๐—ป๐˜ ๐—ฆ๐˜†๐˜€๐˜๐—ฒ๐—บย that acts as an Operations Manager. It doesn't do the work; it delegates. ๐Ÿ”น ๐—ง๐—ฎ๐˜€๐—ธ ๐—”:ย "Find letters that sound grateful." โ†’ Routed to theย ๐—ž๐—ป๐—ผ๐˜„๐—น๐—ฒ๐—ฑ๐—ด๐—ฒ ๐—”๐˜€๐˜€๐—ถ๐˜€๐˜๐—ฎ๐—ป๐˜. ๐Ÿ”น ๐—ง๐—ฎ๐˜€๐—ธ ๐—•:ย "Check the Coal Warning List." โ†’ Routed to theย Ops Agent. The Supervisor combines the intel into one perfect answer. We didn't write any complex routing code, we just defined the team, and Agent Bricks handled the rest. ๐Ÿ—“๏ธ Start Day 9:ย medium.com/startups-at-databโ€ฆ ๐Ÿ”” Missed previous days? Catch all the demo drops here:ย medium.com/startups-at-databโ€ฆ ๐Ÿ’ป Get the Code:ย github.com/arjuncode/12daysoโ€ฆ #Databricks #AgentBricks #MultiAgent
9
Itโ€™sย ๐——๐—ฎ๐˜† ๐Ÿดย of theย ๐Ÿญ๐Ÿฎ ๐——๐—ฎ๐˜†๐˜€ ๐—ผ๐—ณ ๐——๐—ฒ๐—บ๐—ผ๐˜€, and we are solving the "last mile" problem between Santa's data and his AI agents. ๐ŸŽ… Usually, if you want an AI to check "Toy Inventory" or "Reindeer Health," you have to build a custom API layer. You write the SQL, then you write the Python wrapper, then you deploy the container. Itโ€™s heavy, slow, and brittle. Today, we remove the middleware. Usingย ๐——๐—ฎ๐˜๐—ฎ๐—ฏ๐—ฟ๐—ถ๐—ฐ๐—ธ๐˜€ ๐— ๐—ฎ๐—ป๐—ฎ๐—ด๐—ฒ๐—ฑ ๐— ๐—–๐—ฃ (๐— ๐—ผ๐—ฑ๐—ฒ๐—น ๐—–๐—ผ๐—ป๐˜๐—ฒ๐˜…๐˜ ๐—ฃ๐—ฟ๐—ผ๐˜๐—ผ๐—ฐ๐—ผ๐—น), we took 20+ SQL functions and instantly turned them into callable AI tools. We gave the AI "hands" to help run the North Pole, without building a single server. ๐Ÿ—“๏ธ Start Day 8:ย medium.com/startups-at-databโ€ฆ ๐Ÿ”” Missed previous days? Catch all the demo drops here:ย medium.com/startups-at-databโ€ฆ ๐Ÿ’ป Get the Code:ย github.com/arjuncode/12daysoโ€ฆ #Databricks #Startups #AI #ManagedMCP #DatabricksForStartups #MCP
9
๐—ฅ๐—ฒ๐—ณ๐—น๐—ฒ๐—ฐ๐˜๐—ถ๐—ป๐—ด ๐—ผ๐—ป ๐—ฎ๐—ป ๐—ถ๐—ป๐—ฐ๐—ฟ๐—ฒ๐—ฑ๐—ถ๐—ฏ๐—น๐—ฒ ๐—ป๐—ถ๐—ด๐—ต๐˜ ๐—ผ๐—ณ ๐—ฐ๐—ผ๐—ป๐—ป๐—ฒ๐—ฐ๐˜๐—ถ๐—ผ๐—ป ๐—ฎ๐˜ ๐—”๐—ช๐—ฆ ๐—ฟ๐—ฒ:๐—œ๐—ป๐˜ƒ๐—ฒ๐—ป๐˜ ๐—ถ๐—ป ๐—Ÿ๐—ฎ๐˜€ ๐—ฉ๐—ฒ๐—ด๐—ฎ๐˜€!ย ๐Ÿš€ Databricks Ventures, Wipro Ventures, and Capital One Ventures brought together a powerhouse group of founders and VCs to discuss the evolving landscape of the data and AI ecosystem. Grateful to our partners and every founder who took the time to step away from the conference for a night of community and collaboration. #Founders #Startups #DataAI #AWSreInvent #DatabricksForStartups #Databricks
33
โœย ๐—œ๐—ณ ๐—ฎ ๐˜„๐—ถ๐˜€๐—ต ๐—น๐—ถ๐˜€๐˜ ๐—ฐ๐—ผ๐—ป๐˜๐—ฎ๐—ถ๐—ป๐˜€ "๐—ผ๐—ถ๐—น ๐—ฝ๐—ฎ๐—ถ๐—ป๐˜๐˜€", ๐—ฎ ๐—ธ๐—ฒ๐˜†๐˜„๐—ผ๐—ฟ๐—ฑ ๐˜€๐—ฒ๐—ฎ๐—ฟ๐—ฐ๐—ต ๐—ณ๐—ผ๐—ฟ "๐—ฐ๐—ฟ๐—ฒ๐—ฎ๐˜๐—ถ๐˜ƒ๐—ฒ ๐—ด๐—ถ๐—ณ๐˜๐˜€" ๐˜„๐—ถ๐—น๐—น ๐—ฟ๐—ฒ๐˜๐˜‚๐—ฟ๐—ป ๐˜‡๐—ฒ๐—ฟ๐—ผ ๐—ฟ๐—ฒ๐˜€๐˜‚๐—น๐˜๐˜€. That is a data retrieval failure. And itโ€™s exactly why we are upgrading the North Poleโ€™s search engine onย ๐——๐—ฎ๐˜† ๐Ÿณย of ourย ๐Ÿญ๐Ÿฎ ๐——๐—ฎ๐˜†๐˜€ ๐—ผ๐—ณ ๐——๐—ฒ๐—บ๐—ผ๐˜€. Standard text search is brittle. It relies on exact matches. If the user doesn't guess the exact keyword the writer used, the insight is lost. ๐—ง๐—ผ๐—ฑ๐—ฎ๐˜†, ๐˜„๐—ฒ ๐—ฎ๐—ฟ๐—ฒ ๐—ฟ๐—ฒ๐—ฝ๐—น๐—ฎ๐—ฐ๐—ถ๐—ป๐—ด ๐—ธ๐—ฒ๐˜†๐˜„๐—ผ๐—ฟ๐—ฑ๐˜€ ๐˜„๐—ถ๐˜๐—ต ๐——๐—ฎ๐˜๐—ฎ๐—ฏ๐—ฟ๐—ถ๐—ฐ๐—ธ๐˜€ ๐— ๐—ผ๐˜€๐—ฎ๐—ถ๐—ฐ ๐—”๐—œ ๐—ฉ๐—ฒ๐—ฐ๐˜๐—ผ๐—ฟ ๐—ฆ๐—ฒ๐—ฎ๐—ฟ๐—ฐ๐—ต. In this demo, we build a pipeline that automatically converts millions of letters into vector embeddings, allowing Santa to search for concepts. ๐Ÿ—“๏ธ Start Day 7:ย medium.com/startups-at-databโ€ฆ ๐Ÿ”” Missed previous days? Catch all the demo drops here:ย medium.com/startups-at-databโ€ฆ ๐Ÿ’ป Get the Code:ย github.com/arjuncode/12daysoโ€ฆ #Databricks #Startups #VectorSearch #MosaicAI #DatabricksForStartups
3
๐Ÿ’ฌ "๐—›๐—ฒ๐˜† ๐—š๐—ฒ๐—ป๐—ถ๐—ฒ, ๐˜€๐—ต๐—ผ๐˜„ ๐—บ๐—ฒ ๐˜๐—ต๐—ฒ ๐˜๐—ผ๐—ฝ ๐—ด๐—ถ๐—ณ๐˜๐˜€ ๐—ถ๐—ป ๐—ค๐˜‚๐—ฒ๐—ฏ๐—ฒ๐—ฐ." Welcome to Day 6 of the 12 Days of Demos. Today, we stop forcing the logistics team to write SQL, and we start letting them speak their own language. We just deployed @Databricks AI/BI Genie to the North Pole operations floor. Now, instead of filing a ticket and waiting 3 days for a report, an elf can simply ask:ย "๐˜ž๐˜ฉ๐˜ช๐˜ค๐˜ฉ ๐˜ฑ๐˜ณ๐˜ฐ๐˜ท๐˜ช๐˜ฏ๐˜ค๐˜ฆ๐˜ด ๐˜ข๐˜ณ๐˜ฆ ๐˜ณ๐˜ฆ๐˜ฒ๐˜ถ๐˜ฆ๐˜ด๐˜ต๐˜ช๐˜ฏ๐˜จ ๐˜ต๐˜ฉ๐˜ฆ ๐˜ฎ๐˜ฐ๐˜ด๐˜ต ๐˜ฃ๐˜ช๐˜ค๐˜บ๐˜ค๐˜ญ๐˜ฆ๐˜ด?"ย and get a visualization instantly. But is it safe? If an Elf asks:ย "๐˜š๐˜ฉ๐˜ฐ๐˜ธ ๐˜ฎ๐˜ฆ ๐˜ต๐˜ฉ๐˜ฆ ๐˜ณ๐˜ฆ๐˜ข๐˜ญ ๐˜ฏ๐˜ข๐˜ฎ๐˜ฆ๐˜ด ๐˜ฐ๐˜ง ๐˜ต๐˜ฉ๐˜ฆ ๐˜ค๐˜ฉ๐˜ช๐˜ญ๐˜ฅ๐˜ณ๐˜ฆ๐˜ฏ ๐˜ช๐˜ฏ ๐˜›๐˜ฐ๐˜ณ๐˜ฐ๐˜ฏ๐˜ต๐˜ฐ,"ย what happens? Genie refuses. Not because of a prompt instruction like "๐˜—๐˜ญ๐˜ฆ๐˜ข๐˜ด๐˜ฆ ๐˜ฃ๐˜ฆ ๐˜ฏ๐˜ช๐˜ค๐˜ฆ," but because of architecture. Give your business users answers grounded in your data. ๐Ÿ—“๏ธ Start Day 6:ย medium.com/startups-at-databโ€ฆ ๐Ÿ”” Missed previous days? Catch all the demo drops here:ย medium.com/startups-at-databโ€ฆ ๐Ÿ’ป Get the Code:ย github.com/arjuncode/12daysoโ€ฆ #Genie #BusinessIntelligence #ConversationalAnalytics #12DaysOfDemos #Databricks
35
Itโ€™sย ๐——๐—ฎ๐˜† ๐Ÿฑ of theย ๐Ÿญ๐Ÿฎ ๐——๐—ฎ๐˜†๐˜€ ๐—ผ๐—ณ ๐——๐—ฒ๐—บ๐—ผ๐˜€, and we are fixing a problem every data engineer knows too well:ย ๐—Ÿ๐—ผ๐—ด๐—ถ๐—ฐ ๐—ฆ๐—ฝ๐—ฟ๐—ฎ๐˜„๐—น. The North Pole had a "single source of truth" for their data, but they didn't have one for their logic. The code to mask sensitive names or calculate regional metrics is copy-pasted across 50+ different notebooks. If you fix a bug in one notebook, itโ€™s still broken in 49 others. Thatโ€™s not just tech debt, thatโ€™s a security risk. ๐—ง๐—ต๐—ฒ ๐—ฆ๐—ผ๐—น๐˜‚๐˜๐—ถ๐—ผ๐—ป: Unity Catalog Functions. โœ… In todayโ€™s demo, we stop writing code in silos. We elevate our SQL functions to beย ๐—ณ๐—ถ๐—ฟ๐˜€๐˜-๐—ฐ๐—น๐—ฎ๐˜€๐˜€ ๐—ฐ๐—ฎ๐˜๐—ฎ๐—น๐—ผ๐—ด ๐—ฎ๐˜€๐˜€๐—ฒ๐˜๐˜€โ€”governed, permissioned, and versioned just like our tables. ๐Ÿ—“๏ธ ๐—ฆ๐˜๐—ฎ๐—ฟ๐˜ ๐——๐—ฎ๐˜† ๐Ÿฑ:ย medium.com/startups-at-databโ€ฆ ๐Ÿ”” ๐— ๐—ถ๐˜€๐˜€๐—ฒ๐—ฑ ๐—ฝ๐—ฟ๐—ฒ๐˜ƒ๐—ถ๐—ผ๐˜‚๐˜€ ๐—ฑ๐—ฎ๐˜†๐˜€? ๐—–๐—ฎ๐˜๐—ฐ๐—ต ๐—ฎ๐—น๐—น ๐˜๐—ต๐—ฒ ๐—ฑ๐—ฒ๐—บ๐—ผ ๐—ฑ๐—ฟ๐—ผ๐—ฝ๐˜€ ๐—ต๐—ฒ๐—ฟ๐—ฒ:ย medium.com/startups-at-databโ€ฆ ๐Ÿ’ป ๐—š๐—ฒ๐˜ ๐˜๐—ต๐—ฒ ๐—–๐—ผ๐—ฑ๐—ฒ:ย github.com/arjuncode/12daysoโ€ฆ #Databricks #DataGovernance #DataEngineering #UnityCatalog #SQL #12DaysOfDemos
13