- Basic cloud development skills
- An interest in AI at the edge
We set out to make edge infrastructure easy and ended up democratizing Edge AI for everyone.
Edgeworx was started over 4 years ago with the goal of enabling edge computing at a larger scale solving the issues of remote deployment, management, and operations. To do this, we created Eclipse ioFog to enable developers to treat the edge like any other cloud, making it accessible to all developers.
When COVID hit, it became apparent to businesses that they needed a way to operate remotely. We used ioFog as a platform to develop and deploy a solution to help keep businesses open and running. We realized that while ioFog made it easy to deploy edge applications, building AI applications that took advantage of real-time video, audio, thermal, etc. inputs was incredibly challenging. Moreover, developers now needed to have a deep understanding of AI models and frameworks such as TensorFlow. Those people were few and far between, and to find people who understood both edge and AI were unicorns!
We successfully deployed our edge AI solution and continued to refine it in the field. This left us with a clear vision of how to fill the massive gap we discovered and a suite of tools to do it.
We successfully deployed our edge AI solution and continued to refine it in the field. This left us with a clear vision of how to fill the massive gap we discovered and a suite of tools to do it. We had solved the infrastructure problems at the edge, but realized this is just part of the pain for developers. Developing AI is also super hard. Something we saw time and time again with customers and partners. We set out to make edge AI development easier and more accessible. What we came up with was the Darcy product suite: Develop AI applications with Python using the Darcy AI Engine, deploy to the Darcy Cam a sleek, all-in-one mini edge supercomputer, and operate all your applications and devices remotely with Darcy Cloud. Why shouldn’t all cloud developers be able to build and deploy AI applications that process real-world inputs as they occur?
Introducing the Darcy Suite! AI, Cam, and Cloud
Software, hardware and infrastructure for building AI at the edge.
Build: Darcy AI
Empowers all developers to build AI intended for deployment in real-time, in the real world, real fast. Developers can now build AI applications using Python as opposed to having to become TensorFlow (or insert AI framework of choice) experts. They can chain multiple models together to build what we call an AI pipeline, to run multiple AI models in parallel, and rapidly iterate and adjust parameters at runtime to make the AI most performant. To get you started, we provide an out-of-the-box people-centric AI that anyone with Python or web development skills can use.
# Get the person closest to the camera
# Get the cropped face picture of the person in front
Our AI Engine is extensible, so that developers can plug in existing Tensorflow AI models. All of the processing can be done locally, in real-time, on real video input, and developers can store the data locally or send it to any backend. But what developer wants to listen to marketing? Our team have built a sample application called AI Explorer that allows you to do what you really want to do, try it for yourself. Take your edge hardware (or if you don’t have one run a VM in the cloud) and play with a real AI application in minutes.
Read more about Darcy AI and get code examples head here.
Deploy: Darcy Cam
Smart, sleek all-in-one edge supercomputer. Currently to get AI models running at the edge, the hardware is usually an exposed board with a number of attached sensors including a camera for video. To create a product that has everything, companies would have to commit to months of R&D to make an edge device that fits. That is just too much work (trust us, we went through it, so no one else would have to!). Further, we found that to build real business applications at the edge, we needed multiple AI accelerators to run models simultaneously. Because who is just going to do one thing at a time? Additionally, we needed wifi and bluetooth in a small, unobtrusive, low-profile form factor. And just because we are developers, doesn’t mean we don’t want style, so we added some flair to feed our techno lust to come up with Darcy Cam.
- Up and running in 5 minutes, easy add wifi, connect to the cloud and you’re good to go
- Dual coral AI accelerators, video + thermal sensors, wifi + BLE, runs multiple models for blazing real-time performance
- Manufactured and distributed globally by our build partner ASUS
- Join the wait list to get your own!
Run: Darcy Cloud
Operate your remote hardware and software just like the Cloud. Built on top of our open source projects ioFog and deviceplane, we've added a beautifully crafted UI and robust features to enable the management and operation of applications at the edge. Prefer CLI or API? We've got you covered too, with fine grained application token access. It's clear to us, if it's not easy to deploy applications, no one will adopt edge computing. With Darcy Cloud, orchestrating and deploying applications and then managing their full lifecycle just became simple.
- Deploy Applications defined in yaml, through drag and drop or cli, orchestrate containers to your edge device
- Secure keyless SSH access to any of your edge devices. No VPN tunneling, or port forwarding required
- Connect any hardware from a Raspberry Pi0 to a Cray super computer, or any cloud VM
- Simple organization of your edge device and app topologies
- And much, much more
Build AI apps locally, deploy them anywhere, run them from the Cloud
Got a jump on everyone and are already using our Cloud? Great! Just login with your existing account to cloud.darcy.ai.
- Start developing with the Darcy AI Engine
- Learn how to get Darcy AI on your Pi
- Sign up for Darcy Cloud