At Silo.AI, we’ve always had a desire to work on our own R&D to ultimately get good stuff faster to our customers. We had rapid growth last year, and we’ve now reached the point where we can truly commit and start building a serious product-driven business on the side of our primary service and solution business. With that, we’re fired up to build a brand new in-house R&D team to focus on the long-term development of AI-driven products.
In a nutshell, most of our service and solution work is focused on bringing new AI capabilities into existing product development initiatives, covering for example:
- Smart vehicles (incl. automotive, maritime and heavy machinery)
- Smart devices (incl. mobile, telecom and healthcare),
- Smart cities (incl. infrastructure, energy, ports and logistics)
We spent the late last year accumulating customer understanding around our product ideas. Also, we took that time to plan how our devotion to building AI for people could be reflected in products that keep humans central to the AI learning loop. Now with a strong commitment to building the product business and plenty of exciting ideas to build, we’re getting into the action with our own R&D team.
When it comes to AI technology, for Silo.AI it most often means building solutions based on deep neural networks. However, building reliable products with AI, which in the end is all about probabilities, requires huge amounts of top-notch software to ever function or succeed.
AI products are, after all, software products with some sprinkle on top. That’s also why our product R&D is now looking for strong in-house software developers. The developers we’re after might have no previous exposure to AI, but plenty of software experience and desire and ability to quickly gain a high-level intuition on how AI works. We often meet people who see AI as the future, but struggle to find the right spot to enter the field. For developers with their mindset I’d say that this is one of their greatest opportunities to jump on the AI wagon!
What is the problem we’re solving with products?
In contrast to more traditional software, AI as technology is still young enough that some reasons make it harder for companies to really invest in AI. First, there is often little reference history of similar solutions to rely on. Second, there’s some actual technical risk in building “good enough” models to reach business objectives. Productizing AI modules helps both reduce uncertainty around the technology’s potential benefits and reduce costs in deploying the more common AI use cases. And that’s where we see the big potential for our R&D.
On another note, operating AI solutions requires a somewhat different skill set than what’s usually present in IT departments. For example, AI solutions usually consider and treat data as descriptive and abundant rather than factual and structured, and monitoring, debugging, and updating AI differs wildly from the case of common software. Therefore we’re operating some of the production solutions we build in our own cloud. That often means pretty great cost savings over the lifecycle, but now we need to really turn our operating platform to a solid service offering and build R&D around things like access management, solution monitoring, and inspecting and auditing AI decisions.
Building in-house R&D team
We’re starting the product business with five people. There’s me, Lead AI Solution Architect Niko Vuokko, working on the business. In addition, we’re seeking four developers with impressive experience and great communication skills to join us in our Helsinki office. We’ll also be growing the team constantly throughout this year. The team will keep in close touch with the rest of us, our AI experts, our product people, and most importantly, with our customers.
With AI it’s good to keep in mind how strongly the capabilities of the technology directly determine what’s possible in the business. So in our case it’s even more important than usual to have a seamless understanding between R&D and business, but we should have our bases covered quite well on this. I’ve written software for more than a decade with a wide set of modern tech stacks. That helps me to know what’s easy and what’s hard in R&D and give freedom to the technical team to find the right way forward. We pride ourselves in our technical AI expertise in our service business and there’s certainly no intention to change that with our product business.
The tech stack used will definitely contain Python for its strengths in AI, and we also have Node.js and Vue.js running on top of Kubernetes in our existing codebase. Other than that, many of the tech stack, process, and tool decisions will get direction from the R&D team.
I’ll take the risk and sound cheesy: AI is one of the generational tech advances of our time. For now it’s still just replacing logic internals, that is, it’s an ingredient innovation. But for us close to the ground, it’s more than clear that things won’t stay that way for long. Electricity was a bunch of tweet-worthy point solutions at first before taking over the whole society, but it took 100 years to build the infrastructure to get there! AI is only starting its multi-decade ramp into prominence, and we’re holding on tight to keep up with its unstoppable rise.
If you want to chat about the R&D team or the products we’re building, you can reach me here or send me email at email@example.com. Ps. Not sure if you already applied? Join the R&D team at https://silo.ai/careers/#positions !
Join the 5000+ subscribers who read the Silo AI monthly newsletter to be among the first to hear about the latest insights, articles, podcast episodes, webinars, and more.