Garden Introduces BLOOM for Real-Time IP-Aware AI Drug Design

Garden has launched BLOOM, a Markush structure search engine designed to provide AI drug-design teams with near-instant verification of small-molecule IP landscapes, announced in a press release. This tool allows researchers to iterate on candidates with legal certainty built into the process, addressing the bottleneck of diligence in AI-generated molecule design.

BLOOM employs a graph-based, agentic traversal to compare Markush queries against millions of SMILES strings, effectively short-circuiting invalid candidates through local atom and bond features. The system provides color-coded mapping to confirm atom- and bond-level compliance, transforming verification from a manual task into an automated step.

In benchmark tests, BLOOM demonstrated a 32.44× speed improvement over standard core-extraction string searches, while also reducing false positives that often occur with legacy methods. The engine integrates with Garden's patent database, linking every SMILES match to underlying patent records and supporting workflows from rapid novelty triage to freedom-to-operate analysis.

According to Adi Sidapara, founder and CEO of Garden, BLOOM enables IP-aware exploration without slowing down discovery, ensuring that AI-generated chemistries are checked for existing IP coverage instantly.