Scrap Reduction and Operational Efficiency for the Metal Industry
Why we invested in Tvarit
Matterwave engages in many ways with specialised industrial solutions, and for the most, investing in a company digitalising the die casting market is the apex of specialisation. But if we take a look a bit longer than a cursory glance, we will find that the metal casting market and specifically die casting is fundamental to the industrial sector, allowing complex metal components to be made in series. The metal and machinery used, represent a large part of operational costs for industrial companies, so digitalising and optimising this process represents a true opportunity for cutting resource costs, CO2 emissions, and recovering lost revenue. The die casting market is still lagging behind in digitalisation and data readiness, and many players are struggling with capably collecting and utilising their data. In short, definitely not a “move fast and break things” market. An average of 5%, which represents USD 4bn, of global revenue, is lost annually due to inefficient processes. Making the market, which is currently under significant external stresses (think electrical vehicles gaining in popularity, rising ESG regulations, and geopolitical tensions and Covid-19 affecting supply chains), ripe for disruption and perfectly poised for a data-driven solution to increase resource efficiency.
As we have seen over the past months and years, lots of companies started to realise the opportunities that were presenting themselves in the under-digitalised manufacturing sector. We saw many companies in our dealflow offering basic predictive maintenance and quality inspection solutions, enabled by IoT sensors and some AI/ML models. After some time, we came across Tvarit, a startup from Frankfurt am Main focusing on process optimisation for die casting, which really stood out to us. They had a truly unique offering of an industry-specific solution, and seemed (and we know now, really are!) committed to providing high-quality and maximum-accuracy quality predictions based on their operational model.
Tvarit’s product offering is centred around the die-casting process, focusing on reducing scrap rates and improving operational efficiency. Their big-picture goal is to expand outside of die-casting into other metal manufacturing processes, but for the past three years they have focused on providing a laser-accurate model for die-casting. This accuracy is enabled by a portfolio of more than 160 AI models that build on top of each other to optimise production output, energy usage, and machine maintenance processes…How? By predicting product quality based on deterministic and derived machine parameters, and then validating these predictions with physical models. This allows quality deviations to be detected before they occur, and humanly explained via root-cause analyses. Tvarit took the rigorous approach of creating this hybrid model of physical and AI models, and targeted it at detection, prediction, and explanation. By tailoring its product offering to the physical processes of the metal industry, specifically the die-casting industry, Tvarit was able to achieve a highly accurate modelling representation of processes, leading to a clear value-add with understandable and trustworthy results. In the ultimate understandable terms: rejects are reduced by 40–60%, plant owners can save over 1m EUR, and up to 7.2 GJ of energy consumption (equivalent to ~860kg of CO2 emissions) per plant.
This high-level explanation of Tvarit’s product does not completely convey the uniqueness of their technology. The clear distinction between them, and other process optimisation companies, is to hide the complexity and refinement of their modelling behind explainable recommendations for parameter adjustments, and a broad and approachable platform that intuitively supports the user to leverage the hybrid model technology. The many inputs of material data, domain know-how, machine and quality control data, are all combined into multiple models: Metallurgy-process specific AI models of phase transfer equations, finite-element methods, self-learning AI models, and fine-tuned bespoke AI models for every different industrial context. A head-spinning combination that covers all aspects of the die-casting process, making their product better than the common predictive-maintenance and quality control black-box model. This combination of modelling parts may be overwhelming when listed out but combines to a holistic process optimisation approach that is easy-to-use, integrated into existing workflows, and clearly actionable and explainable. There is proof to these statements: Tvarit’s customers include Maxion, the world’s largest wheel manufacturer, Kamax and other leading players in this industry. Tvarit has been able to generate recurring revenues since its first fully commercial year and is in a strong position to capitalise on the strong pipeline of new customer lead and ongoing pilot projects.
While the above factors all speak strongly for Tvarit, the truly convincing factor was the TEAM. It was their extensive domain experience and true commitment to their customers that made them stand out among the many process optimisation companies we saw. The founding team is led by Suhas Patel and Rahul Prajapat. Suhas gained his significant experience through positions at Qualcomm and Intel and has since dedicated himself to entrepreneurship. Rahul acted as a Network Design Engineer at NTT Communications for Industry 4.0 implementations at Mitsubishi, Hyundai, and Toyota, and worked on a Mars Rover project with NASA. The two founders have worked together since 2017, and since 2019 as and for Tvarit. Together, the team leverages and develops its deep domain expertise by working closely with their customers to really understand the pain points across the industry and to build and deliver highly accurate and applicable solutions. Apparently, others agree with us, as they were selected as one of the best 8 out of 490 AI companies across Europe in 2020 (EDI) and were awarded at #1 place with the DIGITAL FUTUREcongress Start-Up-Award 2022. We are looking forward to working together with this truly outstanding and committed team over the next years to realise their vision of dramatically increasing digitalisation and resource efficiency in the metal industry.
For more information about Tvarit, check out their website: https://www.tvarit.com/
Feel free to reach out to us, Robert Gallenberger and Victor Szabo, if you are interested to learn more about our view on the market or if you would like to get in touch with the founders.
