Everyday AI Explained, Part 1: Our 5 Core Tenets

June 27, 2018 · 3 min read
29654

Many companies are experimenting with AI solutions, but few are actually putting them into production. Why is this? Well, running a single AI model to create some interesting, even if not impactful, results, can be a challenge in itself. So what happens when 10 models are needed? Or 1,000? WorkFusion has customers that run upwards of 25,000 models.

The point is, implementing AI in-house is no small feat. So how do the big AI players pull it off?

Google leveraged its considerable resources to build its AI operations from scratch, from servers to processors to its proprietary software platform. Amazon, Facebook and Netflix also have their own top-to-bottom AI stacks. Microsoft even shuttered its Windows division — which had endured since 1980 — to focus on AI. We know of 19 major companies that are now following in Google’s footsteps and building their own AI chips.

But your company (and we mean this in the best way possible) is not Google. An everyday organization like yours isn’t going to spend billions building its own chips and proprietary AI stacks. That’s why WorkFusion’s Smart Process Automation (SPA) software offers Everyday AI.

What is Everyday AI?

Everyday AI is a focus on simple, self-service enterprise solutions you can implement in days, not years. It’s about moving beyond the theoretical and aspirational, and instead, offering you what’s practical and possible. It’s about providing software powerful enough for data scientists, but easy enough for business users and developers to pick up as well. No need to build your own chips. No need to build your own stack. We take care of that, so you can focus on the data and its impact on your organization.

Above all, in order to work for you, we believe Everyday AI should be:

1. Manageable

An AI is only as good as its data sources, so it’s important that you have complete governance over yours. Our solutions make it easy to route data in and out. They also keep track of your various models and their numerous versions, so you can keep building without the need for a dedicated data science team.

2. Performant

Your industry moves fast, so your AI models need to work fast. Our built-in Process AutoML is designed to train quickly on small data sets. And since you probably don’t have a fleet of GPUs at your disposal, it’s also built to run in your existing environments, on the infrastructure you already have.

3. Measurable

Statistics may tell you “how much” or “how many,” but a snapshot of the past is not enough to make a dent on your business’s future. With our Analytics features, your models will yield results, not stats. They’ll predict outcomes, and offer up new knowledge you can use to truly move the needle.

4. Resilient

Like all the smartest humans you know, a good AI should be able to handle mistakes, and admit fault when they happen. Thanks to the AutoQC processes built into our SPA platform, your models can test themselves for errors and alert you as soon as issues or unexpected changes occur.

5. Continuous

Your business is continuously evolving, so it makes sense that your AI should be continuously learning. AutoQC helps your models analyze exceptions and tweak their behavior accordingly. Plus, it offers plenty of tools for fine-tuning, so you can bounce back from errors without missing a beat.

By injecting these tenets of Everyday AI into SPA — our breakthrough automation platform — we hope to make artificial intelligence accessible to more than just the Googles and Facebooks of the world, but to any data-driven enterprise that’s looking to start making an impact with AI. Every day.


Also published on Medium.

Drive productivity with Intelligent Automation Cloud
Drive productivity with Intelligent Automation Cloud