Garden has launched an innovative tool called BLOOM that transforms how researchers verify the intellectual property (IP) status of new molecules. This new search engine can check small-molecule IP landscapes almost instantly, saving valuable time in drug development. As AI models now generate molecules at a rapid pace, ensuring these molecules don’t infringe on existing patents has become a major bottleneck. BLOOM aims to eliminate that delay by offering fast, automated IP verification that keeps up with AI-driven discovery.

Fast and Accurate IP Verification with BLOOM

BLOOM uses a unique graph-based traversal method to compare complex chemical structures against millions of patent records. It quickly processes Markush queries—generalized representations of chemical groups—by matching them against vast databases of SMILES strings, a standard way to encode molecules digitally. The system produces color-coded maps that show atom-by-atom and bond-by-bond compliance, making it easy to see if a molecule is already protected by a patent.

This approach turns what used to be a manual, time-consuming task into an automated process. In testing, BLOOM was shown to be over 32 times faster than traditional string search methods, completing comparisons in just a fraction of a millisecond. It also identified relevant matches that older methods might miss, especially when subtle bond differences are involved. This speed and accuracy help researchers make faster decisions during the ideation phase, avoiding costly patent infringement issues later on.

Seamless Integration and Practical Benefits

BLOOM is integrated with Garden’s comprehensive patent database, giving users instant access to underlying patent documents. This allows AI agents to not only check for existing patents but also to summarize, compare, and prune molecule sets based on IP considerations. Researchers can now conduct rapid freedom-to-operate analyses and patent landscape scans directly alongside their molecule design workflows.

According to Garden’s CEO, Adi Sidapara, the tool closes the loop between molecule generation and patent awareness. He explains that while AI can propose thousands of potential chemistries quickly, BLOOM instantly reveals which of those are already protected. Kavin Sivakumar, a lead ML researcher at Garden, adds that BLOOM’s ability to capture small but critical changes around molecular groups helps identify patentability quickly. This means design work no longer needs to be slowed down by lengthy patent searches or false positives that waste time.

The impact of BLOOM is potentially huge. By streamlining the patent checking process, researchers can focus on creating high-quality candidates rather than worrying about patent conflicts. It accelerates the entire drug discovery pipeline and reduces the risk of infringing on existing IP, making innovation faster and more efficient.

Inspired by

• https://ai-techpark.com/garden-launches-bloom-for-real-time-ai-drug-design-ip-checks/