MongoDB Ep 141: Building Smarter Apps with AI
How to Make Your App and Database Understand Human Language With One Line of Code
Here's a summary of the podcast we prepared using One AI's Language Studio:
One AI is a Language AI platform dedicated to help developers build smarter apps using AI. The platform is leveraging MongoDB among other technologies to serve its advanced NLP APIs.
Amit believes Language AI should be a fundamental, platform-based solution, just like any other underlying technology.
What is One AI?
One AI is an NLP platform for developers to process audio, video, and text. You can convert it to structured data you can then plug into your product and database to create value for your customers.
One AI has tens of thousands developers registered on the platform using the language studio, which is a low code environment designed to experiment with and discover different capabilities, and auto-generate code snippets to embed in their projects.
What are the Skills that you're providing?
Language Skills are packaged NLP models that provide value for different domains and use cases. So for example, there's an Emotion Detection Skill. You can plug any type of input, whether it's textual, spoken, Word or blog posts, and get the emotional metadata of that text.
Language Analytics Engine
The One AI Language Analytics engine allows you to move from processing a single object such as a document, podcast, blog post or a tweet to processing millions of objects.
How Does the Platform Work?
There are multiple high level components in the stack. Some of it is the deployment environment that holds all of the machine learning capabilities. For every Skill, there's an underlying set of models. So for every one of those, the platform engineering team prescribes the right combination of resources and the pod that is required to run it. There is another level of logic in the Kubernetes deployment that combines and makes sure all of these pods are being deployed in the right set.
Why One AI is a Great Choice for Developers?
A core tenet of our platform is that behind each Skill, there's not one model. There's an assorted model collection that is trained for different domains. So our Skills are trained on multiple domains and automatically choose the right flavor that will provide the best results.
Listen to the full podcast.