cmendez

Hey HN! Datagran Co-founder here. We first introduced ourselves a few years back with what we thought was a cool idea: mixing workflow automation with machine learning, kinda like zapier.com for ML. At the time, it was a shot in the dark, but many got excited with the idea because we were abstracting the complexities of putting AI models into production.

Fast forward to today, and after many hours talking to clients, we’ve honed in on what really matters: developers want to build their own data infrastructure, but the real problem was in laying it out for the business folks. How can we make data not just available, but also understandable for them where and when they need it? That’s the puzzle we’ve been piecing together.

With that in mind, we brought together all of our past experience in regards to serving data quicker and easier and arrived at a solution that we think will be extremely useful: Datagran APPs.

With our platform, developers will be able to create Apps with the power of AI + functional programming—enabling them to build applications like HR churn management tools, develop data models, create intelligent AI bots, and send data to any destination in minutes.

In a nutshell, Datagran makes it easy for developers to deliver internal data tools, empowering business users to query their data as needed, whenever they need it—including through NLP capabilities.

Think of Datagran as the AI-powered alternative to Retool.

PS: Apps is in Beta and we will add more components as we see usage.

Thanks!

In Episode 1, learn how to connect data, clean it, and perform the first analysis—all powered by AI!

Like and subscribe for more episodes!

Created a script which ultimate objective is to automatically create table docs in Notion to then feed a Slack data bot to ensure correct output. Step by step: -Automatically fetches data from a PostgreDB and creates docs in Notion assisted by OpenAI. -Fetches the documentation from Notion and feeds a Slack Bot to ensure correct output.

Benefits: - Automatic docs created in Notion. - Helps bots to correctly query the data based on accurate documentation.

Fully managed Real-Time Machine Learning Model infraestructure as a service.

This App is the Roam Research for media storage. It uses the VGG Face pre trained database and TensorFlow to detect faces. It also uses Apple's NLP and Google's Image Recognition. It is still a work in progress but you will get the idea in the demo.

OMG I was waiting for this for so long. Congrats!

Co-founder here. We’ve been working on Datagran for the last 2 years and day and night for the last 3 months. We’ve waited to show the product to this community until we thought we had something at least we feel proud of. So after many iterations, months of ups and downs, tears and joys, here it goes...

During my first endeavor it was time to set up our data ecosystem. We started integrating Segment, Stripe, Google Cloud, DataStudio, Heap and ClearBrain. It took us about 2 months to set up everything, and it never worked the way we expected. Data was never accurate, every data iteration was an eternal process and everytime we needed to run a prediction we had to download and upload data with the added difficulty of figuring out how to send the output to a business application.

We thought then if it wouldn’t it be nice to have one place where we could just connect our customer and behavioral data, aggregate and dedup it, run custom SQLs, run a couple of ML algorithms when needed and just send the result to a business application like Google Sheets, or a Chatbot. All in the same environment without having to worry about setup, stichting everything up, uploading and downloading data, among many others.

What we did then was to build an all-in-one data workspace, 5 tools in 1, that help companies integrate data, visualize, run ML pipelines and take action all in one place. Think of Datagran as like Zapier for ML. Making it so instead of companies needing both ML expertise and product expertise to ship an ML pipeline, they need neither. With a simple editor any person can now connect sources, select an operator (can be a Custom SQL or an algorithm) and then an action to send the results to a business application that is embedded in our system.

Now if you need to predict churn for example and send the output to let’s say your chatbot, you can just connect Datagran’s SDK, connect to Stripe, join your data, run a ML algorithm and send your output to Twilio or wherever you want.

In this link https://app.datagran.io/bda/coreOS/workspaces/create you can find a playground of our Beta. Just log in and select the Take a Tour option. It will load a bunch of data and examples for you. The idea in a couple of months is to build more than 20 different Spark algorithms and more actions. We are also building lots of templates with our team of data scientists (just like Webflow does it) for the non technical folks. We will have templates for Business, Marketing, Sales, among others and the idea is to sell those Templates and give data scientists a percentage of the sale.

Hope you like it and would love your feedback.