1. IPRally is a growing patent-search platform provider servicing global enterprises, IP law firms, and multiple national patent and trademark offices.
2. The company uses machine learning and natural language processing to transform over 120 million global patent documents into document-level knowledge graphs for faster and more accurate search results.
3. IPRally built a customized ML platform using Google Kubernetes Engine and Ray to balance efficiency, performance, and streamline machine learning operations, saving 70% of R&D costs with Spot instances and closing a €10m A round investment to continue growing and improving its AI platform for patent searching.
IPRally, a patent-search platform provider, is experiencing rapid growth as it services global enterprises, IP law firms, and patent and trademark offices around the world. To meet the increasing demand for accurate and efficient patent searches, IPRally has trained its models for greater accuracy, adding 200,000 searchable records for customer access weekly and mapping new patents.
Using a graph-based approach, IPRally has transformed the text from over 120 million global patent documents into document-level knowledge graphs embedded in a searchable vector space. This allows patent researchers to receive relevant results in seconds with AI-selected highlights of key information and explainable results. To support these efforts, IPRally built a customized ML platform using Google Kubernetes Engine (GKE) and Ray, an open-source ML framework, to balance efficiency, performance, and streamlined machine learning operations (MLOps).
The combination of GKE and Ray provides IPRally with the scalability and performance needed for complex training and serving needs, as well as the ability to efficiently scale down capacity when not in use. Leveraging NVIDIA GPU Spot instances within GKE optimizes operational costs, allowing IPRally to manage compute resources effectively. A custom orchestration layer, IPRay, further simplifies MLOps by providing a command-line tool for data scientists to provision Ray clusters easily.
By utilizing Google Cloud and open-source frameworks, IPRally has been able to build an enterprise-grade ML platform without significant costs. The company has saved 70% of ML R&D costs by using Spot instances and closed a €10m A round investment recently. With a focus on improving its graph neural network models and AI platform for patent searching, IPRally is poised for further growth in the intellectual property industry.