Interview with Citrine Informatics: AI, Data, and the Future of Materials Science

What Sets Citrine Apart

 

According to Greg, four factors distinguish Citrine from others working in the space:

 

• Ease of use: The platform is designed to be transparent and intuitive so that scientists can directly pursue their material science objectives.
• IP model: Unlike competitors who attempt a pharma-like model of discovering and patenting new compounds, Citrine does not create or claim ownership of materials IP. Customers own their discoveries entirely.
• Focus on experimental data: While DFT and molecular dynamics (MD) are supported, the platform is centred on experimental data at all scales—from molecules to macrostructures like paints, plastics, and composites.
• Optimized use of small data: Recognizing that experimental datasets are often limited, the company has built AI models that are chemistry-aware and extract maximum insights from relatively small datasets.

 

This portfolio-based approach allows Citrine to work across many industries, rather than being locked into a single materials class or application.

 

IP Considerations in AI-Driven Materials Research

 

One of the most frequent topics raised by customers is intellectual property. Greg highlighted four key concerns:

 

1. Data security: Ensuring confidential data stays protected, encrypted, and never shared with others.
2. Cross-learning risk: Customers worry about whether their inputs might feed into models used by competitors. Citrine’s strict contractual and compliance rules prevent this.
3. External databases: Companies often want to leverage published materials databases, but these can be of limited or ambiguous value. Citrine helps customers integrate external data selectively where relevant.
4. Inventorship in the age of AI: With debates about whether AI systems can be inventors, Citrine advises that it should always be the scientist steering the AI who is listed as the inventor. AI narrows possibilities; humans decide and prove which candidates become real inventions.

 

Citrine maintains a clear division: the company owns and develops the software IP, while customers own their materials IP and know-how.

Challenges at the AI–Materials Intersection

 

Integrating AI into materials research isn’t just a technical challenge; it requires cultural and organizational change. Greg identified three main hurdles:

 

• Data quality and relevance: Many companies assume decades of legacy data will be invaluable, but much of it turns out not to be actionable. Citrine emphasizes starting with modern, well-structured data collection and then carefully targeting older datasets.
• Cultural resistance: Senior scientists who built careers without digital tools may perceive AI-driven experimentation guidance as a challenge to their expertise. Early wins with supportive teams help drive cultural transformation.
• Connecting with business leadership: Many chemical companies lack processes for procuring digital tools, and business leaders can be slow to recognize R&D digitization as a key driver of competitive advantage.

 

What’s Next in AI for Materials

 

Greg is most excited about the next phase and key challenge of data availability: unlike language models that train on vast internet-scale datasets, materials data remains limited. Success depends on representing chemistry effectively and training on small, high-quality experimental datasets. This creates opportunities for shared models, where pre-training could be done collectively—with layers then fine-tuned on proprietary data.

 

He also envisions future academic and industrial collaborations adopting new “data value metrics”, analogous to citation indices, to credit researchers not just for publications but for the datasets they contribute to the community.

 

A Digital Future for Materials Discovery

 

From its roots in Excel-driven experiments to today’s chemistry-aware AI models, Citrine Informatics has been at the forefront of evolving how material scientists work with data. With unique positioning around usability, data security, and a business model aligned with customer IP ownership, Citrine continues to push the boundary of what AI can bring to the materials industry.

As Greg reflects, the goal has always been straightforward: enable material scientists to express their expertise digitally, and empower them with the tools to turn that expertise into innovation.

 

For tailored guidance on intellectual property at the intersection of AI and materials science, please get in touch with Materials and Software specialist Monica Patel (Monica.Patel@Keltie.com). Our dedicated AI and Materials Practice brings particular expertise in navigating the unique IP challenges and opportunities that arise where advanced materials innovation meets AI-driven discovery.