Now loading...
BASF has taken a significant step in optimizing its global supply chains by developing a digital twin of its complex network using AlphaEvolve. According to Dr. Goetz Krabbe, Vice President for Global Supply Chain at BASF, previous attempts relying on deterministic models had not succeeded. However, the incorporation of AlphaEvolve now enables the company to accurately map its network based on system data while capturing the human decision-making processes that impact daily operations. This advancement allows for a data-driven digital twin this is both precise and manageable, enhancing planning and forecasting capabilities by over 80%. For further insights on BASF’s implementation of AlphaEvolve, visit their blog.
Coolblue has successfully leveraged AlphaEvolve to improve its e-commerce demand forecasting significantly. The company’s data scientists focused on automating feature engineering and model selection, resulting in more than a 5% improvement in production forecasts within just 200 iterations. This improvement was achieved through an ensemble of regression models and refined target preprocessing. Accurate forecasting is critical for managing stock levels in both the short and long term, and AlphaEvolve has proven instrumental in optimizing bulk purchasing decisions and ensuring stock availability. For more about Coolblue’s advancements, check out their website.
FM Logistic has refined its warehouse routing strategies, reporting a remarkable 10.4% enhancement in efficiency due to the integration of AlphaEvolve and Gemini in partnership with Google Cloud. Rodolphe Bey, Group CIO at FM Logistic, noted that this improvement builds upon an already optimized baseline, resulting in quicker order fulfillment and less strain on operational resources. To learn more about FM Logistic’s successes, visit their blog.
Infineon has initiated the use of AlphaEvolve in its chip design processes, observing promising results that could revolutionize the chip development lifecycle. Michael Kollig, CIO of Infineon, highlighted its potential for enhancing various developmental stages, including surrogate modeling.
JetBrains is also using AlphaEvolve to improve the performance of its Integrated Development Environment (IDE). Dmitrii Batkovich, Director of Engineering at JetBrains, described how AlphaEvolve simplifies performance optimization tasks that were previously too intricate or time-consuming, allowing engineers to focus more on the critical aspects of their work. For insights into their IDE performance enhancements, visit JetBrains’ blog.
Kinaxis has made notable strides in its forecasting and optimization systems using AlphaEvolve, achieving over a 22% improvement in forecasting accuracy metrics while significantly cutting runtime by 90% on benchmark datasets. Gelu Ticala, Chief Technology Officer at Kinaxis, emphasized the importance of such advancements for effectively managing increasingly complicated supply chains. Further details can be found on their website.
Klarna employed AlphaEvolve to enhance the throughput of its largest machine learning training pipelines, successfully doubling its output while also improving the quality of the models under strict regulatory requirements. The engineering team at Klarna highlighted that the system autonomously explored nearly 6,000 candidate architectures, leading to innovative deep architectural rewrites that human engineers might not have considered. You can explore more about Klarna’s experiences with AlphaEvolve in their blog.
Kuro Games is applying AlphaEvolve to enhance its backend server performance, aligning well with its philosophy that AI should improve the quality of work, not just its speed. Lin Chenchen, CTO at Kuro Games, noted substantial gains in specific server-side workloads due to this optimization approach.
In a significant collaboration with Google DeepMind, Oak Ridge National Laboratory has deployed AlphaEvolve on the Frontier supercomputer to optimize GPU kernel development for exascale computing. Oscar Hernandez Mendoza, a Senior Computer Scientist at ORNL, mentioned that the partnership allowed for an exploration of optimization candidates that traditional methods might have overlooked, marking a potential milestone in scientific software improvement.
Old Dominion University has taken on a complex challenge in modeling biological aging mortality rates using AlphaEvolve. The Qin Lab successfully rediscovered established models and demonstrated significant improvements in their composite fitness scores, paving the way for future research into multi-species datasets. Dr. Hong Qin of Old Dominion University emphasized the robustness of the models evolving from diverse structures. Further insights can be found on their website.
PacBio has enhanced the accuracy and cost-effectiveness of its genomic sequencing instruments using AlphaEvolve, positioning researchers to uncover disease-causing mutations that were previously challenging to detect. Aaron Wenger, Senior Director at PacBio, pointed out the significance of this improved data quality for scientific discovery. More details on PacBio’s advancements can be found in their latest blog.
Pebble has successfully tackled GPU performance modeling through AlphaEvolve, achieving a dramatic reduction in model errors. Keval Shah, Head of AI at Pebble, described how AlphaEvolve’s autonomous discovery of performance modeling formulations has optimized their inference serving capabilities. The company looks forward to using these advancements for continuous improvements.
Qbraid has made significant progress towards enhancing quantum computing efficiency with AlphaEvolve, yielding results that surpass years of manual research. Kenny Heitritter, Vice President of Research and Development at Qbraid, noted that AI-driven systems like AlphaEvolve would accelerate the development of practical quantum technologies. For a deeper dive into this topic, check their blog.
Schrödinger reported that AlphaEvolve has expedited molecular simulations in drug discovery, contributing to faster research and development cycles by allowing quicker screening of molecular candidates. Gabriel Marques, ML Tech Lead at Schrödinger, stated that this advancement enables the screening process to occur within days rather than months. For more about their breakthroughs, visit their blog.
Substrate has used AlphaEvolve to optimize its computational lithography frameworks, achieving remarkable increases in runtime speed and efficiency. James Proud, CEO of Substrate, expressed excitement at the potential for these models to design future iterations of themselves. More information on their developments can be found on the Substrate blog.
WPP has turned to AlphaEvolve to improve its digital marketing campaigns by significantly improving the accuracy of performance predictions. Anastasios Tsourtis, Lead Data Scientist at WPP, highlighted the success of employing AlphaEvolve to explore candidate model architectures autonomously, resulting in notable increases in prediction accuracy and downstream recommendation scores. For further details, refer to their blog.
Beyond partnering with a high number of companies, Google is also integrating AlphaEvolve into its infrastructure to improve its models. As highlighted by Pushmeet Kohli from Google DeepMind, AI is evolving from merely acting as productivity tools to becoming powerful engines of discovery. The use of AlphaEvolve has improved operational efficiency in various projects and held promise for future advancements across a high number of fields.
Organizations interested in beginning their journey with AlphaEvolve will find it accessible with just two inputs: a seed program, which designs the initial algorithm, and an evaluator that scores adjustments to optimize performance. This simpler setup allows seamless integration into existing workflows for transformative machine learning outcomes.
