Skyflow CEO on Data Privacy with Databricks at AWS re:Invent

Summary

• Data intelligence from unstructured data: Databricks enables deriving insights from receipts, photos, and contracts not captured in tables.
• Data privacy wrapper: Skyflow helps businesses manage personal data by providing a privacy wrapper around Databricks for secure handling.
• Unified platform advantage: Standardizing on Databricks lets organizations build data warehouses, clean rooms, and AI applications in one place.

Skyflow CEO on Data Privacy with Databricks at AWS re:Invent

Try Databricks today: https://dbricks.co/3EAWLK6
Anshu Sharma, CEO of Skyflow, shares strategies for embracing generative AI and data privacy using Databricks on AWS. Learn actionable tips for starting your AI journey. #DataPrivacy #GenerativeAI #AWSreInvent

Chapters

• 0:00 Introduction
• 0:21 What is Skyflow
• 1:40 What is Data Intelligence
• 2:40 Real Time
• 3:34 Data Preparation

FAQs

What is data intelligence as described in this video?

Data intelligence means deriving insights from all types of data—structured tables, unstructured files, photos, receipts—enabling analysis on day one rather than months of processing.

What problem does Skyflow solve?

Skyflow helps businesses manage personal data they receive (phone numbers, credit cards, dates of birth) by providing a data privacy wrapper that ensures compliance and security.

Why is Databricks important for data privacy?

Databricks provides a unified, managed, and open platform where organizations can handle large information volumes in one place, improving security by reducing need for multiple systems.

What are the advantages of standardizing on Databricks?

Standardizing on Databricks allows organizations to build data warehouses, clean rooms, and generative AI applications in a single unified platform rather than integrating five components.

Full transcript

0:00 Introduction

Welcome to the data bricks experience at Cafe Grand Lux Venetian at AWS reinvent 2024 so by way of quick introduction I'm with Anu Sharma co-founder and CEO of skyflow Welcome thank you could you tell us a little bit about skylow and what your thinking about these days so skyflow is a company that's focused on only one thing uh helping make sure that when we give our personal data to businesses so imagine you're filing your taxes you go to coffee shop actually these days you can't even hail a cab without sharing your phone number credit card number date of birth so as a result businesses are forced to handle a lot of information as they're building their apps as they're building their data LS or they're building gen applications including agents we help them manage all of that with our skyflow data privacy W what are AWS and data break's role in skylow strategy so datab brakes has really changed the game by building a layer of data that's both managed and open is how I think about it and as a result we see a lot of companies basically standardizing on platforms like data brakes to build their data warehouses their clean rooms their gen and as you do do that you end up with a lot of information in one place right and what's magical for datab bricks customers is they can all do it in one place they don't have to go to five different components and get that working so what does data intelligence mean to you to skyflow I love the idea that now for the first time really in the last 30 years we can derive information from data that may not be already structured so we've always been able to go into a database table or a spreadsheet and say hey tell me how many apples you bought yesterday but if that's not captured in a table but it's there in your receipts it's in your photos it's in your files it's distributed in some contracts and I want to ask you know I don't a large insurance company I want to say hey how many contracts do we have which might be at risk well it may take weeks months years for some of the companies to go through pedabytes of data but with data intelligence to me means is that intelligence is available on top of the data from day one so you can ask intelligent questions off of your data that may not be already predicted well that kind of thing is only possible with a data intelligence platform and not with a classical data platform how does Real Time come into play how important is that for some of the capabilities you mentioned you know I spend my entire career in the data space I was at Oracle for seven years I've moved to Salesforce where another Salesforce platform and there was always this big divide between what I can find out about the last few months versus what I can find out about what's happening today or yesterday but really platforms like data Bricks now that have combined the speed at which they can ingest and index the data so when you go to a coffee shop or you go to a Walmart store or you go to some other place they should be able to figure out maybe what your next need is and maybe give you a coupon or an offer and all of that is only possible if the system can work in near time or real time and so to me the power of a system like data brakes is they make that information available to you in near real time so you can actually build those apps so when you're working with customers and they want to embark on one of these Aid driven projects and build out those agents what would you advise for them to evaluate before the work begins that there's actually going to be payoff at the end so I think most PE companies don't have the data in the shape and form that really is usable so you know we've all used chat GPD it's good for the first three four answers but then when you get into and you want to ask about yourself or in the case of businesses I really want to help you figure out which flight to rebook well that's no longer a common sense question I call them Wikipedia questions like it's not something you can train a model on Wikipedia and rebook your flight I need to know more about you do you fly business class do you fly private do you fly you know do you have TSA Pre you know do you like rebooking a flight next day or you usually take the next flight available all of that information has to be first made available in a form that these agents and AI can actually consume so whether you call it data preparation we call it you know getting the data in the right shape it includes things like data prep data cleaning you know things like security and privacy controls sometimes you're sanitizing the data anonymizing the data if need be if you're training or fine tuning the models all of that work is really data work Brick by Brick and if you do that correctly you're very nicely set up to start building and playing with your first generation of AI apps which I think is mostly what we call play mode and play mode often is going to be prototypes you probably have trusted people playing with it but when you from move from play to production all the work you've done with the data on control security privacy sovereignity is going to be critical otherwise we're seeing a lot of companies have very successful demos and prototypes and when they try to roll it out either at their customers if you're a startup or you're internal and you're rolling it out to your employees or customers you start seeing the holes right because you're basically letting anybody in any language ask any question of all of your data like just imagine that's like a hacker's dream come true so we think you have to prepare the data Breck by Brick put security and privacy controls in place around your pi and other types of sensitive information and that's step one well onu thank you for flow's partnership with datab bricks and sitting down to chat with me today thank you so much

Learn more about the Databricks Data and AI platform.