Informatics & AI for Materials Development
Justine Trubey, Citrine Informatics
While the generation of 3D geometry using AI is a relatively new endeavor, AI and ML for material chemistry has been used in practice at industrial scale for a number of years.
Citrine Informatics is a leading software platform that leverages AI and their data management infrastructure to help some of the biggest materials companies on the planet transform their material development process to produce better performing and more sustainable materials, chemicals and products.
Justine Trubey, Chief Operating Officer of Citrine Informatics will be discussing at CDFAM how Citrine’s AI empowered platform helped large companies use their own data to accelerate their material development process to help create a better, more sustainable future.
Could you begin by providing a brief overview of Citrine Informatics, its mission in the materials science and AI industry, and some examples of the types of applications it has been used for?
The Citrine Platform, a SAAS product, combines domain-specific AI algorithms and smart data management infrastructure to establish structure-property relationships of materials and allows for the storage, retrieval, and interpretation of these data with high fidelity and interoperability. Additionally, the software offers improved data visualization and tools for integrating domain knowledge.
The Citrine platform allows users to leverage AI and MI without needing to develop their own MI infrastructure in-house. It can help users meet sustainability goals through design, significantly accelerate product development, and provide chemicals and materials companies stronger flexibility and adaptivity to handle changing market demands due to megatrends, short-term events, and customer preferences.
The Citrine Platform (CP) accelerates the pace of innovation across a wide range of industries, including specialty chemicals, coatings, batteries, and CPG.
We have a deep sustainability mission to enable more performant and sustainable materials development faster.
Can you give some examples of the types of applications the Citrine Platform and AI-driven Materials Informatics has been used for?
Here are just a few, ranging from the mundane to the sublime (anonymized):
- Taking PFAS out of an company’s entire supply chain
- Developing a lighter metal to improve energy efficiency in airplane engines
- Creating shampoo that doesn’t separate on the shelf
- Choosing the most performant solvent blend (with complex ingredients and processing parameters) from more than 850 million options
- Optimizing organic semiconductors for IOT applications
- Making a cheaper and better non-slip coating for tools
What motivated you to originally join Citrine Informatics? Could you also shed some light on how your experiences at organizations such as Random House, Amazon, and Shapeways have contributed to your decision to join and shaped your approach as the COO?
I like to make stuff. Optimization is my happy place.
My roles at Random House and Amazon were about implementing technology to solve optimization problems and enable massive transformation in traditional manufacturing and content distribution.
Citrine’s mission is to help notoriously slow moving industries find more performant, and sustainable materials and transform the way they make and dispose of them.
Shapeways was at the crossroads of all the biggest next-gen tech solutions, enabling short run, customization, on demand, and democratization. 3DP and generative design have literally changed our concepts of design and production at the most basic level: what does a “hole” looks like, and how would you make one has an entirely different set of answers.
I joined Citrine because we have the chance to change the world by using cutting edge technology to get dangerous chemicals out of our bodies, soil, water, and air. The global manufacturing industry is roughly $10 trillion, with a spectacularly urgent need for optimization.
During your time at Shapeways, the company was focused on utilizing industrial 3D printers, which were essentially prototypes themselves, for mass production on a 24/7 basis. Can you share specific insights or lessons you gained from that experience which have been valuable in your current role at Citrine Informatics?
Shapeways had a brilliant concept to change the world by making manufacturing available directly to consumers by pushing the boundaries of an emergent machine tech.
Shapeways had a brilliant concept to change the world by making manufacturing available directly to consumers by pushing the boundaries of an emergent machine tech. I joined in 2012 and Shapeways had demand but couldn’t consistently deliver good quality products.
We built a global supply chain of all the 3DP processes in every possible material, and set out to push the limits of those testy experimental machines in 24/7 production of high-mix low-volume.
Materials were constrained: We had what the machine manufacturers gave us, and we relentlessly fiddled with laser settings, temp controls, tray planning, post processing to make good quality parts, but we never had any control over the materials we used.
Until I met Citrine, I never thought to ask “What would happen to the quality of my parts if I had a material with a different flow rate?”
Lessons learned from Shapeways:
- 3DP for the masses is not all that compelling to the masses. Most people would rather just buy their stuff. The compelling opportunity is in “industry.”
- No matter how much you tweak a machine or a process, the solution lies in the combination of process, machine, and materials.
- Digitization looks the same everywhere, from making books, mass customized jewelry, and 1-off industrial parts. I’ve never worked in a lab inventing materials, but I have been on the shop floor smacking my hand against my head trying to make stuff with the wrong materials, where every moment of machine downtime and poor quality output wastes tens of thousands of dollars.
Can you provide us with an overview of the key topics and themes that you plan to cover in your upcoming presentation at CDFAM, “Informatics and AI for Materials Development”?
- How our customers perceive materials in their strategies, the problems it solves and how that all changes with AI and data driven methodologies
- What is Materials Informatics and Sequential Learning, and why digitize product development?
- Accelerated Product Development
- Data Driven R&D Portfolio
- Enterprise Knowledge Management
- What does “better ma
- Using data in the product development and mfg space
- Competitive advantage
- Manufacturing efficiencies
- Sustainability and why hasn’t every company done this yet?: Obstacles
- The Grey Areas
- Some cool examples
For companies that may face cultural friction when adopting new technologies like AI, what steps does Citrine Informatics take to help these organizations overcome any potential resistance and successfully integrate AI-driven solutions into their operations?
Think about the first days of Salesforce or SAP: everyone must be bought in, projects managed, figure out how to change what you have done forever, be given the space and prioritization to accomplish it. Digitization and Data-Driven production needs to be a priority up and down the organization.
The Investment must be aligned with a strategy and also be attached to a business model and business outcomes.
This generally means leaders must see a huge potential ROI and competitive advantage, while also appeal to a few levels of constituents:
- CTOs want to understand what AI can do for product strategy, new offerings, better pricing, and beating the competition
- P&L owners, need to buy the competitive advantage, the financial ROI, so they can prioritize the extensive digital transformation and cultural shift across labs, supply chain, procurement, IT, etc
- Scientists and manufacturing engineers need to believe that data-driven materials development will make their jobs better, rather than making them obsolete, and they need to believe the results are from real science and not just luck
Citrine helps build the ROI story so companies can get to HOW and WHY
- Increasing throughput of innovation
- Decreasing formulation complexity and inventory “bloat”
- Reducing costs
- of existing products, including supply chain disruptions
- of manufacturing by making materials a variable on the shop floor (see above about Shapeways!)
- Addressing regulatory requirements to remove toxicity from their portfolio of products or the manufacturing process
Could you elaborate on how Citrine’s AI-driven approach can enhance the agility of materials exploration and development in order to respond effectively to sudden changes in supply chain dynamics or regulatory environments?
Citrine came to a virtual standstill during the first few months of Covid. No one went to the lab, ingredients were not being mined or made, and lines were shut down. There were epic supply chain disruptions, massive cuts in manufacturing capacity, and unprecedented, unpredictable changes in demand.
Most manufacturers struggled to get raw materials; but once the first kick of dust cleared, they needed to go into overdrive on production. With China (and other logistics solutions) almost completely out of commission, the raw materials that manufacturers always used to make chemicals and products were no longer available, and they needed to find new ones.
(Thankfully) Regulatory pressures are becoming more stringent, every day brings a declaration about bans or thresholds in manufacturing. Companies need different formulations, new suppliers, and different processes to make and throw away new products, with optimized outcomes.
In many cases, the lack of high-quality data can be a significant obstacle to adopting AI in optimizing manufacturing processes. Does Citrine Informatics face this challenge, and if so, how does the company overcome it to ensure accurate and effective AI-driven solutions?
Citrine’s approach doesn’t require huge data sets–we’ve had very good results with a hundred lines of data. That said, most companies have thousands, if not hundreds of thousands, of data points from R&D alone (well into the millions if you added shop-floor), but therein lies the greatest opportunity, and the greatest obstacle, for these companies
It can be a huge undertaking to make it useful for an informatics approach. Our customers have excel files, fox pro databases, LIMs systems, and even books of handwritten results with data that all use different abbreviations and notation protocols, without process mapping, and sometimes it’s missing altogether, having retired along with the guy who left 10 years ago. There must be a commitment at the top levels and throughout the teams to prioritize it.
We offer tools and services that can clean up data, but more often the biggest obstacle is gathering it from across the company, and giving someone air time in their workday (prioritizing and making it a critical expectation) to get it done.
When working with sensitive information from clients and partners, how does Citrine Informatics ensure data privacy and security to prevent cross-pollination of information between clients?
Materials data is at the center of our customers’ IP and their strategic competitive advantage, so security is first priority to them, and also to Citrine.
We are ISO 27001 certified, so our InfoSec policies are scrutinized and we have passed with flying colors. In the end, it’s not enough to have controls, we needed to build an entire culture of information security with our Citrine team. We have done that successfully and continuously reinforce it, and every person at Citrine treats information with the confidentiality, care, and respect it deserves.
How can a company determine if they have a materials-related problem that Citrine Informatics can help solve, what are the indicators or signs they should look for? If they do, what types of data, infrastructure, and human resources are necessary for companies to effectively adopt and implement Citrine’s materials informatics solutions?
Materials are purpose-designed with properties that serve specific use cases, and business success in materials development is founded on the ability to strategically capitalize on optimization opportunities.
Every company has a business need for developing or improving materials or products:
- Better manufacturing and supply chain
- Selling a new product to a new customer to keep up with tech developments
- Respond to regulatory pressures
- Create a distinguishing property among thousands of competitors
- Deliver “boutique” materials, and…
Here are some concrete examples of a materials problem:
- You have 300 competitors for the paint you make, and you are struggling to get to the top of the shelf.
- You are using additive manufacturing to develop airplane engine parts, and the lighter equals more cost efficient and sustainable so you can sell kajillions of them.
- You make adhesives and your customers start worrying about the toxicity of offgassing.
- You use a readily available chemical in your polymer but current geo-political forces have tripled the price and lead time.
- You are producing products that contain Pfas and you know it’s killing the earth.
- You need to stop using lead and mercury in your manufacturing process because it can make your workers sick, and there is no good way to dispose of it so you are killing the planet.
Finally, what do you hope to take away from your interactions with other experts at CDFAM exploring AI in their work but may not actually be doing so yet. Do you have any words of caution or advice for them?
The way we consider design and manufacturing is entirely upended by digitization. It is cross-functional, global, and affects every single aspect of our industry and human development.
AI itself is a juggernaut–one of the most powerful developments in our lifetimes. I am excited to hear what my colleagues are doing, learn about the grey areas between good and evil, and where we are headed.
To hear more more from Justine and other AI and manufacturing expert’s fight against evil, register to attend CDFAM to ensure you are also on the right side of history.