Our recent webinar with Modern Synthesis raised an interesting point. Ioannis Zampetakis, Head of Science at Modern Synthesis cited a 30% increase in speeds of development after just over four months of a project applying the Alchemite machine learning method. He went on to say that they expect these gains to increase as the technology beds-in. But what was notable was the focus on speed-of-impact.
A key corporate goal for Modern Synthesis is to scale-up the processes associated with their innovative biomaterials, moving from pilot to full-scale production within a year. When project timescales are short, a new approach – like using Alchemite machine learning to inform experimental design – is only useful if you see its benefits quickly.
But there is another reason why a fast start matters when introducing a new technology. Momentum builds confidence and drives adoption. A slow start can be a project killer, even when the project seems to be showing potential for long-term gain. Introducing a new technology or modifying research approaches can be difficult. Researchers are only human. Making even small changes to workflows requires effort and mindshare that it can be hard to dedicate, given the need to continually deliver on day-to-day objectives. Too many innovation projects fail simply because they don’t show enough early progress.
At Modern Synthesis, Ioannis Zampetakis reported that the research team – like any group of good scientists – was initially “curious but not convinced”. What got the project moving was focusing on a particular experimental recommendation from Alchemite that “went through our labs and gave us a huge step improvement in our development. Within a month, we were able to get this first unlock.“
What can deliver such a fast impact? Obviously, as we’ve discussed on this blog before, when adopting new software, the user experience is vital. Tools need to slot easily into the tasks and workflows they are intended to support. But the structure of and support for a project matter too. Finding a problem where you can run a quick, impactful pilot project is a great way to get started. And the right coaching also helps. In the Modern Synthesis case, the Intellegens Science Team worked closely with the internal project team in the early stages. Weekly joint project meetings helped users get their data into the system, identified productive routes for inquiry, and enabled the team to move at pace to generate valuable insights. The aim was to ensure early success while transferring knowledge and skills that will enable self-sufficient use of the machine learning in future.
You can read many more case studies of machine learning on this website. FUCHS reported a 50% performance improvement in performance of a lubricant. Voestalpine cited 40% cost savings in manufacturing and testing of alloys. Johnson Matthey identified routes to a new catalyst formulation that require five times fewer experiments. In general at Intellegens we expect a 50-80% reduction in experimental workloads due to the implementation of Alchemite machine learning. But it’s important to remember that it isn’t just this potential that matters – it’s how quickly you begin to see returns. We’ve developed expertise in the service and support that can deliver this fast impact – if you’re starting your machine learning journey, let’s talk!
