What’s better than an AI/ML app? A platform that can help you integrate machine learning into your existing apps and workflows with a few clicks and a couple lines of code.
Today marks the public launch of Zorroa’s new no-code/low-code machine learning (ML) integration platform, Boon AI, which enables ML automations in visual media management through simplified API-based integrations. What does all of that even mean? It means Boon AI is the easiest and fastest way for product and development teams to integrate computer vision capabilities into their DAM/MAM/production apps using ML APIs from providers including Google Cloud, AWS, and Azure. And that product, Boon AI, is now generally available.
How did we get here?
We’ve come a long way since 2014, and so has the role of machine learning. While “AI” and “ML” have become quintessential terms at any exec-level roundtable on “process transformation,” the market is still struggling to stand up successful ML projects. And for many, ML integrations are still out of reach, as we wrote in our blog post at the end of 2020.
And so we pivoted. The team spent the past year building a SaaS platform that could enable organizations of all sizes to integrate ML at a fraction of the time, cost, and resources required in traditional data science projects. We envisioned our customers building an ML pipeline in a manner that resembled the agility of software development.
“There’s a gap and a need for agile ML that gives media technology teams the opportunity to quickly stand up ML projects at a lower cost of entry, then experiment, iterate, and scale with reduced risk levels. That’s the kind of thing that empowers the acceleration of ML innovations. The Boon AI platform does just that—it provides teams with the tools to quickly build and scale their ML without forcing them to change their fundamental media management workflows,”Marc Stevens, President and CEO of Zorroa
What exactly is the Boon AI platform?
- Single interface for accessing ML APIs from Google Cloud, AWS, and Azure
- Infrastructure to ingest, process, visualize, store, index, and search ML results at scale
- Automation of media management tasks using ML modules for object and label detection, image classification, optical character recognition, and speech-to-text
- Framework for an open source content search app and third party app plugins (e.g. Adobe Premier)
- Ability to integrate ML predictions into customer apps using Python SDK and REST API
The platform focuses on visual media assets including images, videos, and PDF documents. Customers can tap into supported ML APIs to start automating manual media management tasks, including metadata tagging, similarity search, logo detection, product classification, speech transcription, and explicit content detection. Don’t have a media app? Check out our open source content search application.
What ML challenges is Boon AI solving for?
ML implementation surfaces issues more specific than the popular list of AI challenges like data access, talent shortage, and lack of organizational support. When the rubber meets the road, teams discover that it’s not feasible to build something that’s flexible enough to work across multiple use cases, it’s difficult to modify or scale without breaking something in the process, or that vendor orchestration and ML API integration can be a months-long process requiring home-grown tools that not only needed to be built, but also continually maintained.
Limits in traditional AI/ML integration
Organizations investing in AI/ML have traditionally had few options outside of building bespoke, difficult-to-scale ML solutions in-house, hiring expensive integrators, or shoehorning their workflows to fit ML vendor requirements.
The Modern AI/ML integration alternatives
Boon AI flips that concept around by giving technologists the control to experiment, evaluate, then integrate ML results into apps they’re already using through a no-code/low-code solution. No need for ML vendor integrations, custom tooling, or months of development cycles.
Custom vs. off-the-shelf, pre-trained models
A custom-built ML solution is surely the right approach if the task at hand is to build a global fleet of self-driving trucks or a cashier-less checkout system, both of which require the business to solve for a wide range of edge cases and deliver results at near 100% confidence level. But not all computer vision applications require 12-24 months of R&D, 99% accuracy, or built-from-the-ground-up ML models. Many use cases can be solved for by a growing ecosystem of ML APIs available through major cloud ML providers—built, trained, and tested by the best data scientists the tech world has to offer. And if these pre-trained models can solve for a sizable chunk of your ML needs, why not take advantage of them to get a head start?
What’s in a name?
As an API-first company, we built a product that is a first of its kind to break down the AI/ML adoption barrier by enabling ML API implementation with minimal code. To mark its release and our new direction, we are launching the product under the new name, Boon AI. Why “Boon”? It reflects our vision of putting our customers in a favorable position by enabling discovery and exploration through accessible ML.
We’re thrilled to announce a new chapter of our company and our new vision forward.