At Rayven, we’ve worked with lots of partners to craft and deploy IoT solutions and Industry 4.0 applications (using our world-leading Dynamix integrated data, AI + IoT platform) for many organizations across numerous sectors. Through that process, we’ve learned what does and doesn’t work.
It might sound obvious but the key thing in all successful AI + IoT projects is data. However it’s not about MORE data, it’s about the RIGHT data.
Data is needed to not just analyze and build a business case, but to continually feed into any solution built to allow it to operate – without it, everything else fails. In the world of ‘Big Data’, it’s easy to capture everything and get lost in the mess – the true skill is knowing what’s relevant and honing in on that. As you consider the different steps outlined below and apply it to your organization, always have half a mind as to the RIGHT data that you need to collect, how and where it can be stored and analyzed, as well as what’s needed to enable you to take action on it.
10 steps to successful AI + IoT deployment:
- Identify a problem or something that you want to improve in your organization. You may have goals that you’re working towards, be wasting far too many resources, or needing to up output – often this decision has already been made for you because the problems are obvious and growing, so it has begun without AI + IoT even being mentioned!
- Pick a champion. To drive change within any business, you need someone that is willing to educate key stakeholders and champion the project widely. Ideally, they need to be P&L focussed and have both a strategic understanding of what can be achieved, as well as be interested in innovation within your sector. It will be their responsibility to not only make the business case, but then also bring the wider workforce along on the journey so that the project doesn’t stumble post-deployment – change management can be the hardest piece.
- Explore the use case. Consider how AI + IoT technology might be used in your own environments in different scenarios to achieve your overarching goal. For example, if it was to improve efficiency, there’s lots of inputs that go into that, e.g. what’s going on on the factory floor, with your HVAC system, transport, etc. By choosing the right use case, rather than a technology and its supposed benefits, you’ll be able to plan an effective, successful project.
- Analyze what resources are available. It’s likely that there will be internal skill sets or existing technology roadmaps that might be of use, so this is the time to analyze how far they can get you and where you’re going to need to seek further support. A basic understanding of the technologies that will address your use case, as well as what a full AI + IoT stack looks like, is essential and an idea of what the partner ecosystem looks like would definitely help.
- Find the right AI + IoT suppliers(s). Work with organizations that know you, your sector and that have a trusted network of partners who they can call in to craft a complete solution. Anyone that’s ramming AI + IoT down your throats at the beginning, or who is promising a silver bullet, is advisable to be given a wider berth. Whoever you choose to work with, it’s critical that they have an understanding of your industry and that they can translate what you’re trying to achieve into something manageable – also ensure that they also have an eye for the long-term and aren’t offering siloed non-expandable technologies.
- Pick a key metric and look to fill data gaps. Pick what the key metric of success will be and then establish where you have data gaps or the inability to measure. What would an ideal data set look like? What’s currently missing? Do you have historical data that would be of use? From here you can establish what hardware and software will be needed, as well as how automation and business logic models would work in your particular business.
- Identify your technology stack. You should now know not just what you’re looking to achieve, but the data needed to make it run and pieces of hardware that can actually do it. Much like bricks of lego, an AI + IoT technology stack could be made of any number of things – at Rayven we see 5 distinct pieces – and one-size doesn’t always fit all.
- Start small. Run a small pilot program to establish what the business case is. This approach reduces risk and necessary upfront investment, making it far more likely that you’ll be able to begin. Take one production line, store, or handful of assets and work with it; monitoring over time to see (and measure) the impact that it’s having – all with an eye on how this can scale. If you’re working with a good provider, as well as using data science to combine data sets and explore the impact that changes will make, they’ll also be looking to add a layer of machine learning so that, as the model runs in real-life, it’s able to identify further changes that will enhance performance.
- Make the business case. From here you can craft the business case with proven data points for expanding and scaling the solution to different sites, lines or products. With clearly identified metrics and establish-able benefits, you’ll find making the business case simple.
- Scale. Once you’ve proven your AI + IoT solution’s worth, it’s now time to capitalize on its benefits. Roll it out to all production lines, factories, and divisions. The more relevant data that you can capture across your business will help you (or your machine learning tool) to improve further – you could even think about adding further inputs and devices. Working with a partner that can work with you over the longer-term and, critically, knows how to scale will shorten future reduce costs, speed-up the deployment process, and reduce further development cycles.
Start small, build business cases, and utilize organizations and partners that not only work together and are experts in their field, but who are asking the right questions.
Speak to us today to find out more and to get some free advice.