Why your creative agency needs to know about Machine Box
Imagine what a boost your pitches to potential customers would get by solving problems using state-of-the-art machine learning. Sprinkling some artificial intelligence in the right place can produce staggering results. It’s a great way to stand out among your competition and win more business.
Building machine learning into your apps can be a daunting prospect; it’s complicated, messy, and you really have to know what you’re doing to get it right. Most agencies don’t have expensive data scientists on staff and cannot spare the time their developers need to figure it all out.
Consuming Cloud APIs is an option but they require you to send customer data elsewhere, they come with limits and can get very expensive at scale.
This high financial cost removes that all-important mashable, hackable nature that can lead to the kinds of innovation that good agencies are famous for.
“You might just want to see what impact a bit of machine learning can have on a project.” — Aaron Edell
Machine Box delivers state-of-the-art machine learning inside Docker containers, wrapped with a “beautiful API” and “the best developer experience of any product I’ve ever used”.
You just spin up a box, and get hacking.
And when it comes time to go to production, it’s easy and (more importantly) affordable.
Solutions include facial detection and recognition, image classification, natural language processing, nudity and NSFW detection, personalisation and recommendation intelligence… all for text, images, videos and unstructured data.
Scroll down for an overview of the kinds of problems you can solve with Machine Box
Get to MVP quickly
One of the design principles behind Machine Box was to enable rapid development. That meant it needs to be effortless to install, but more importantly, trivial to integrate, deploy and maintain.
We put great effort into our developer experience, and have received very encouraging feedback.
“Machine Box has the best developer experience of any product I’ve ever used” — Dave Cheney
It works like this:
- One line of code to download and spin up a box (run a local Docker container)
- Going to http://localhost:8080 gives developers a console where they can learn about the capabilities of the box, try out the features (without writing any code), and even start a conversation with us, the founders, so we can answer any questions
- Developers write code that uses the API (also hosted on localhost) to solve some pretty advanced problems like facial recognition, image classification, nudity detection, natural language processing, and more, on both images and video
- When you’re ready, deploy the docker containers in your favourite place (in the cloud, or on prem — a nice option for customers who care about the privacy of their data)
Sensible pricing
That old freemium trick of getting you hooked on a product before ramping up the price (also called the heroin model) has never really appealed to us — we wanted Machine Box to always feel like it’s delivering more value than what you’re paying.
It’s still early days for us, but companies that sign up today pay a simple subscription — for digital agencies, each of their clients would subscribe — and you get access to every box, to run at whatever scale you need, for a fixed monthly price.
This approach means you not only know how much you’re going to spend each month (rather than trying to guess how many API calls you’re going to make) but that amount doesn’t change as your apps become more successful.
Use cases
You can solve many problems with Machine Box and creative agencies are coming up with new and innovative ways to improve their products and services all the time.
We’re a little blown away by some of the applications so far, and we’d like to share some of them with you.
Facial detection and recognition with Facebox
Facebox lets you recognise faces in images and videos.
You teach Facebox who people are by showing them as little as one example image, so it can automatically recognise them.
You can use this to improve search and SEO by tagging content, use it as part of an authentication system, automatically find videos based on who appears in them, drive engagement with your users by notifying them when they appear in social content, anonymise images by blurring out stranger’s faces.
Classifying images with Tagbox
Tagbox categorises images, giving you human readable tags that describe them as well as allowing you to group by similarity.
You can drive sales on e-commerce sites by suggesting similar looking products. One customer talked about using it to proactively see where their photography was appearing online.
Tagbox is also teachable, so it can learn how to look for content that’s specific to your use cases. With a single example, Tagbox will adapt to classify images of any kind, allowing you to really get creative.
Understand what people mean with Textbox
Textbox provides natural language processing of unstructured textual data such as emails, reviews, tweets, comments, questions, etc.
You can pull out entities from the text, like places, dates, people, numbers and more. You can get a list of keywords that describe the text, as well as the sentiment of each sentence to determine if the text is positive or negative.
You can really improve SEO by automatically generating metadata that accurately describes your own content. You can process terabytes of user generated content to gain valuable insights that would otherwise require humans to sit and read through. One company even talked about using Textbox to better understand their competitors.
What now?
Hopefully you can see the potential of applying machine learning to some of your current projects, or you would like to arm yourself with a little more information for future pitches.
You can either send this post to your development team and ask for their views (please do this, we pride ourselves on making developers happy) — or get in touch and we’ll happily setup a call to chat about it.
Also:
