When applying new AI-related technologies, so much can go wrong. Not only does the tech have to work but it also has to be used and embraced by those it was formulated to help. Meanwhile, people are naturally suspicious of complexity – especially if it comes out of a “black box”. They also have a healthy contempt of overpromising. Perhaps, most importantly: it takes a lot of time and money. And because it’s an innovation, another innovation can pop up overnight to make your solution obsolete. Anything can happen. For instance, just when you think you are getting somewhere, it turns out your algorithm is racist.
In short: it’s easy to get cut when you’re surfing the cutting-edge. And while innovation means making many mistakes in the process, Augury can now say we’ve achieved a certain level of success with our AI. So why us and not others?
Read The White Paper ‘The Power of Augury’s AI’.
1) Access to clean qualitative data
Augury is collecting its own data from the machines we aim to keep functioning at maximum efficiency. And we’re collecting lots of it – with more coming in as we scale. And that’s super important. We all know the old adage: the insights coming out are only as good as the data coming in (there are more profane versions of this adage). This is why our algorithms achieve an accuracy of over 99%. And because it all works so well, we build trust quickly (speedy ROI also helps).
2) Full stack means no interoperability issues
We just snap on our IoT sensors onto factory machines and suck up the data to the cloud where the AI magic happens that provides real-time insights for those working on the ground. And because we are bypassing all other systems, we don’t have to worry about interoperability issues. Just plug and play!
3) Boldly going where no one has gone before
“We are dealing with AI for a very unstructured and unexplored data domain,” says Tal Gurevich, Augury’s Director for AI Products. “While images and texts have already gotten a lot of attention, we are sorting data in the not widely ploughed fields of vibration, magnetic flux and temperature. We have to come to grips with what sort of supervisory and unsupervised outputs we can bring to the training set in order to extract value. For instance, like how medical AI uses doctors to label data and then train the algorithms, we do the same with our vibration analysts. Basically, we are charting out new territory and that makes it super interesting.”
4) Our domain bypasses many of AI’s “issues”
Thankfully, we haven’t had to be overly distracted by typical AI/data issues such as ‘privacy’, ‘regulation’ or ‘bias’. This brings a certain purity and clarity to our AI while also speeding up the process. So, for example, while the medical world is still largely waiting for better patient outcomes thanks to AI, we are already providing that for machines. Bias is also not relevant in our arena (sure, we’re passionate about rotating machines, but we’re also very open to a nice steam trap).
5) It’s all about Hybrid Intelligence not Artificial Intelligence
Human-machine collaboration is key to any successful application of AI. After all, technology should enhance human activity, not replace it. At Augury, we do this by looping in machine reliability expertise during both development and deployment – and thereby creating a loop of continual improvement. Sure, we love our algorithms but we love our vibration analysts even more. And to ensure adoption, strong UX and change management has become part of our package – to ensure that ‘digital transformation’ is not a vague term but a fact.
6) We are solving a real problem
AI is often touted as the solution to all problems. But few organizations actually dive deep enough in figuring out what their core problems actually are. Only by solving what needs to be solved can you have impact. Shallow use cases are useless use cases!
7) We found the right real-world partners
We found the right partners – namely, our most forward-thinking clients – to go next-level with. We’ve always had rather large ambitions and that always involves making rather large mistakes. Hence, we need partners who are willing to take the risks because they understand the larger vision.
8) Scaling on vision: onward to ever newer data sets
We helped create the new category Machine Health. Now with the acquisition of Process Health pioneers Seebo, we are aiming for full Production Health. By looping in ever-more different data sets, we hope to further maximize the effectiveness and efficiency of the full manufacturing process. More data=more impact=more motivation. But yes: this is a work-in-progress, so stay tuned!
9) Success breeds success
Our approach works. We have been very successful – and we have the magic unicorn dust to prove it. As a result, more doors are now opening than closing – while giving us the confidence to move forward.
10) Our work matters – and so does the talent
We saved the best for last…
It used to be that the only AI gig in town with a reasonable salary was helping speed up click-through rates. Now you can find competitive wages helping cure cancer, elevate sustainability efforts, or keep the machines that matter most running. Data science has suddenly become very sexy – like software engineering did a decade or two ago. And like manufacturing, data science is shedding its reputation of being dull and repetitive. By offering interesting work that makes the world a better place, we attract top talent.
In fact, our amazing team is actually the #1 reason why our AI has been so successful. Now you know!