Interview: How AI Is Transforming Coatings R&D—from Data Silos to Faster Innovation with Nick Talken

Before founding Albert Invent, you led the digital transformation of R&D at Henkel. What were the biggest lessons you learned there that shaped your vision for Albert Invent?

One of the biggest lessons I learned which helped shape everything we build at Albert Invent is that digital transformation is as much about people, workflows, and data as AI itself. I learned firsthand that AI is only as valuable as the quality and accessibility of underlying R&D data. The large chemical companies we work with are amazing and world-class at chemistry, but were not designed for creating the data foundation to enable AI.

The process for standardizing data and workflows has become an essential element of how we approach customers at Albert Invent; we know that for this to be a long-term value driver, the true unlock is when you apply AI to structured data. Running a single experiment to develop a product using a legacy workflow can take weeks or months and be very manual and very slow. It’s expensive, and the way you discover new information is very analog.

Adopting a data-first approach means change management is just as important as technology creation and adoption. Chemists want to play a role in shaping how their work is evolving and understand how to adapt to new tools and technologies. Having conversations with the chemists at the bench and my own experience in that role helped shape Albert’s platform-first approach rather than building standalone AI tools. This eventually mapped to Henkel evolving from paper-based processes toward highly digitized R&D workflows and the success they are seeing with Albert Invent. And we continue to see great results: Henkel has increased speed to market 2x for 3,000 scientists across 142 labs using Albert OS (operating system).

Many coatings companies have accumulated decades of formulation and testing data, but much of it remains difficult to use. What are the biggest barriers preventing companies from turning that historical data into a competitive advantage?

The challenge for many of these companies built before digitalization and AI is the data, not the model. At storied R&D companies which are protecting highly sensitive trade data and patented formulas, historical information often exists across paper notebooks, spreadsheets, disconnected databases, and different sites and business units. Files and archival records documenting years of experiments–they are not on the internet or even a desktop.

Apart from access, this means data lacks standardization and consistency; the fact is, different scientists also record experiments differently. A recent example is one of our customers using their automated lab; their adhesives testing was not as standardized in the past due to the ad hoc method of sample submission, but now through the use of Tasks, Data Templates, and Parameter Groups in Albert OS, their testing data collection has been streamlined to enable the technology and better operations.

Before the model can be adapted to a specific customer there are phases of deployment: upload the unstructured data, in parallel, standardize and structure data so they can get more value out of AI. It’s a meticulous process but once historical knowledge is searchable and structured, it becomes a strategic asset that informs the next generation of invention and innovation at that company.

---

Event tip:

The EC Conference Digitalisation in Coatings Formulation is the ideal platform to learn about the latest advances and trends in the field of digitally assisted coatings production processes. Join experts, researchers and industry leaders as they share their insights and expertise in automation, big data, AI, and predictive modelling to enhance efficiency, accuracy, and speed in formulation, testing, and quality control.

---

AI is often presented as a tool that can replace traditional R&D processes. In your view, where does AI genuinely create value for formulation scientists, and where is human expertise still irreplaceable?

There are so many ways AI can enable the best scientists to be even better at their work, and having legacy knowledge and expertise at their fingertips in a digital format is the starting point. The goal is increasing innovation capacity.

AI can do this through recommending the highest-value next experiment if the first didn’t work, and quickly searching enormous formulation spaces. It can also find patterns humans would likely miss. This all contributes to improving throughput and shortening development cycles. A chemist in Germany isn’t going to run the same experiment their American colleague did last week if they have the data record and know the polymer didn’t work. Our goal is to reduce the time per iteration and collapse it down to as little as possible to help scientists achieve their goals.

This is really about augmenting human innovation. The scientist is telling the machine where to look and how to optimize directionally and constrain the space of possibility, using their expertise and applying chemical intuition. They are the ones who truly understand customer requirements and are able to interpret and make recommendations for unexpected experimental results.

A great example of this is our work with Advanced Polymer Coatings (APC) which has reported approximately 50% faster invention across projects using Albert. The scientists are still making the calls and greenlighting projects moving forward; AI is helping them do this in a way that means they spend more time on high-value innovation.

Sustainability targets are forcing formulators to replace raw materials more frequently than ever before. How can AI help coatings companies identify viable alternatives while reducing development time and technical risk?

We knew when we began building Albert Invent that the materials and chemistry space was already working on the challenge of making more sustainable and less toxic materials. AI offers an enormous opportunity to cut down on waste from unnecessary experiments and also look for better, more sustainable materials at an accelerated pace.

The challenge of sustainability introduces many new formulation constraints simultaneously, and to meet some of these very ambitious and important goals, AI can help manage some complex tradeoffs many companies are making, including product performance, cost, regulatory requirements, and of course manufacturability. Because it’s one thing to design a product in a lab and test it–then you need to be able to source the materials and build the supply chain to bring it to market.

Rather than testing hundreds of combinations, AI can perform simulations and then surface the most promising experiments to test physically. The historical formulation knowledge we were discussing earlier becomes even more valuable during raw-material substitution, as scientists have a more complete picture of what has worked previously. Eventually, this combination translates to faster evaluation, lower technical risk, and quicker commercialization (or go-to-market)

Looking ahead five years, what do you think the “AI-powered coatings laboratory” will look like?

We think the AI chemistry lab of the future will consist of a fully connected digital R&D environment where experiments are automatically captured and AI is continuously learning from historical results, customized to each customer’s unique chemistry.

The goal is to free scientists from admin work, chasing down historical data and spending precious lab time on low-value tasks like reformulation. This will enable chemists to focus on creating the next true breakthrough innovations. We see greater integration between software, automation, and laboratory equipment and faster movement from idea to validated formulation.

We see the scientists leading the charge on defining customer problems and creative formulation strategies, while exercising their scientific judgement particularly when it comes to interpreting ambiguous results and innovation driven by experience and intuition. Essentially, a technology-enabled laboratory which empowers scientists with better information, better recommendations, and dramatically faster learning.

• Save article in bookmarks
• Print this article