At the third and final showcase at VentureBeat’s Transform 2019 AI event, five companies — Skymind, Techsee, EngineML, Brainworks, and Unravel — took the stage to tell their stories and share their latest news.
Machine learning is capable of solving problems and expediting tasks for the enterprise, and Skymind’s goal is to help companies deploy it faster and overcome some of the technical challenges of handling data. The company’s current ML platform runs in the cloud, on premises, and on the edge. With the ability to analyze images, text, and more, it’s found traction in industries like telecommunications, banking, and manufacturing.
With the introduction of Pathmind, Skymind is adding reinforcement learning to its quiver of cloud-based capabilities. It can use simulations and virtual models to provide the best possible decision paths for real-world business needs, for everything from supply chain to traffic control.
“Current solutions that optimize in simulated environments can’t handle the complexity because they don’t offer enough compute and can’t leverage reinforcement learning. That limits the scenarios that they can be applied to,” the company told VentureBeat.
TechSee uses AR and AI to make customer call support visual, intelligent, and engaging. When a customer calls for help setting up a new device, they can use their phone to give the agent a visual look at the issue. Using computer vision, the agent can identify and annotate what the customer needs to do and walk them through it.
With its new TechSee Smart, the company has added a layer of speed to the process. It can automatically identify the issue at hand and route the call to the right agent or self-service channel, and in the case of the former, it can help guide the agent towards the correct problem resolution.
Engine ML’s mission is to provide a way for customers to expedite their machine learning engineering cycles. It offers a SaaS platform that gives customers access to the compute power of dozens or hundreds of GPUs and promises to significantly speed up the training process.
On the Transform stage, they ran a demo where they trained a model that normally takes hours in five minutes. They provisioned GPU resources on the cloud in real time — 64 GPUs on eight cloud hosts.
Engine ML also provides a dashboard so users can track their projects and stay on top of key metrics like Tensorboard outputs and model checkpoints.
Brainworks wants to use intelligent ambient biometric sensors to make healthcare better. The impetus of the idea is that healthcare should be more preventative than reactive. The company’s technology is contactless and always-on, so although it’s constantly checking patients for emotional and physical states, it’s completely physically non-invasive. It uses computer vision-based biometrics, like real-time pulse rate and respiration, to check a patient’s status, and it can even check against a system to ensure that the patient has given consent to be scanned. The technology then checks for FDA-required measurement precision, and if it passes muster, it will automatically add the data to the patient’s record.
This helps doctors better track their patients’ health between visits, and it can help them reach remote patients who can’t make an in-person visit.
Big data is only getting bigger, and much of the challenges inherent to AI have to do with how to manage that data with distributed applications. Unravel provides an AI Ops chatbot that connects to a number of sources to pull in the data it needs for its AI to work. It covers workloads running both on and off-premises. Fresh off of a $35 million funding round, Unravel showed from the Transform stage how it integrates with Slack to learn what you need to do with your application — enhance performance, figure out a cost, and so on — and can then run an AI to perform the work you need or tell you how you can improve it.