How Artificial Intelligence Will Change the Chemical Industry

My hypothesis as a chemist and a follow-up to my 2024 prediction

One of my two big predictions for 2024 is that artificial intelligence (AI) is going to start to really change things for the chemicals sector—it sometimes feels like it’s stuck in the 2002. It’s just that I think it will be through easy stuff that isn’t sexy (like AI directed materials discovery). In the next year we will see a whole host of tools become available to chemists and people within the industry that will enable them to become even better at the 5 jobs they are all currently doing.

Spend enough time working in industrial chemicals across enough end markets and at different companies and you start to see two types of companies out there:

1. Companies that take all generated data very seriously and maintain 20+% EBITDA margins in a very competitive business landscape.

2. Companies that don’t care about you, your data, or anything except boosting revenue and profits right now because we (the executive team and therefore everyone else) are in an emergency and the board just threatened to fire us (the executive team) because we haven’t done much over the last five years to address the continuous decline in top-line revenue and eroding margins (we blamed that on the weather and supply chain issues) that were once so large that they covered up the fact that we didn’t care about anything (like building a viable company with an actual quality system) except making that money machine go BRRRR.

Ironically, I’ve primarily only worked for variants of the second type of company so I might be a little bit biased. I think it’s never been easier to be the first type of company with a whole variety of software companies out there providing ways to document everything from customer interactions to scheduling and inventory management to R&D data. If you really want to do it you can keep track of all your data. The problem is what do you do with it once you have it?

Customer relationship management software (CRM):

1. Salesforce

2. Hubspot

3. [Insert your favorite CRM SaaS company here]

Artificial intelligence tools are already available in these types of companies. It’s just a matter of time before high quality data and reduction of inputs makes things even easier. High quality transcripts and summaries of Zoom meetings with customers over a year paired with generative AI looking at all of the captured data and providing trends that are both obvious an non-obvious is just a matter of time. I suspect either this year or next we will see enormous productivity gains from commercial teams (sales, marketing, etc).

On Enterprise Resource Planning (ERP) or Inventory, Scheduling, & Quality:

1. SAP

2. Oracle

3. [Insert your favorite ERP SaaS company here]

Artificial intelligence tools are already available for SAP, but I think this will be the year that chemical companies actually start to use them. Automating processes usually run by humans like forecasting, picking raw materials via first in first out (FIFO), scheduling, QC data, and whatever else you can think about. I think executive teams will smell a potential for efficiency, cost savings, and a way to reduce headcount in exchange for paying a monthly fee for software. I think the implied promise of this software, which has historically been clunky, annoying and frustrating, will get exponentially better over the next 5 years.

On using the data you already have and will generate:

1. Benchling

2. SciNote

3. [Insert your favorite LIMS SaaS company here]

Everyone is expecting AI and Machine Learning to blow the doors off of R&D and will find/discover stuff that humans haven’t even thought about. Sure, it might happen in the next 10-15 years, but I think the easiest lift here is just being better with the data you already have. A lot of companies struggle with even keeping track of their own data, which is where companies like Benchling and SciNote provide their service of making all of the data generated easy to find and read. That way, R&D teams don’t need to “re-invent the wheel” every five years when the headcount turns over due to poor management. I expect AI tools to look kind of like how Benchling is making them look here, but it could be as simple as a chemist being able to generate a report on all of the work previously done on sustainable feed stocks in the last ten years.

Even just user generated AI GPTs (via Nikolaus Mackay) could make something as nebulous as Rheology for most scientists easier to access and understand. There might even be a future where a rheology n00b like me could do some non-linear experiments. In 2024 we will see a boatload of tools out there being generated. It’s just a matter of which ones are effective and who can use them to their greatest potential.

Custom AI solutions for R&D:

1. Citrine Informatics

2. Chemify

3. [Insert your favorite start-up here]

I’ve saved this category for last because I think it’s maybe the most misunderstood and over hyped. I don’t think that AI or machine learning will replace R&D functions. The big benefit here will be gains in productivity for R&D organizations to scope out potential opportunities of research, scaling up products, or to train models on very specific types of systems with proprietary data. Everything above is more tangible and maybe easier for a finance person to predict such as spending $100-1000 per person per year yields 2-5x productivity. I suspect that the large companies specialized in very specific polymers either are or will be thinking about training a model to better understand their specific area of expertise. In this category I think of the big rubber companies, the big polyurethane system houses, specialty polyesters, olefin producers, making their leading positions even stronger and more difficult to disrupt.

And if they are not doing this then they should get started or some early career scientists are going to figure out how to do it.

Conclusion

I don’t think AI is going to outright replace entire job functions. I think at a minimum it’s just going to make whoever is working in this industry more productive. As the saying goes, “AI isn’t going to take your job. Someone that knows how to use it will.” Now is the time to figure out how to use this stuff.

Let me know in the comments what you think or if you have an AI start-up focused on chemicals or materials that you think I should interview.

Tony