Daily News · 2 min read

Google AI Updates: October 16, 2026

google google-deepmind antigravity protnlm drug-discovery bigquery lakehouse gemini adk agents

1. Google DeepMind Uses Antigravity to Find an Antimicrobial Peptide Candidate Now in Wet-Lab Testing

Google. Google DeepMind published a walkthrough of biologists using Google Antigravity as an agentic science workbench to search for new antimicrobial peptides. The agent queries UniProt entries annotated by ProtNLM, a model that predicts protein function from amino acid sequences, then checks novelty with BLAST against Swiss-Prot and compares predicted structures from the AlphaFold Protein Structure Database, with the researcher steering each step. The session surfaced a 97-amino-acid bacteriocin candidate from the Lactococcin 972 family with predicted activity against pathogens involved in respiratory tract infections, and that lead is now undergoing wet-lab validation. Source

2. Gemini Gets Direct Access to Borderless Lakehouse Data, Including Databricks and Snowflake Catalogs

Google. Google Cloud added a Gemini integration with its borderless Lakehouse, in preview, that lets users set up access to sources such as Databricks Unity Catalog and Snowflake Horizon through catalog federation directly from Gemini. Business users can query cross-cloud records in natural language, and agents can ground reasoning and generate code against the same data without moving it. Several Lakehouse features are now generally available, including Cross-Cloud Interconnect and caching for AWS and Azure, BigQuery support for Apache Iceberg V3, read-write Iceberg interoperability with Managed Service for Apache Spark, and zero-copy integration with Confluent Tableflow. Source

3. google/adk-recipes Replaces adk-samples as the Source for Agents CLI and Agent Garden

Google. Google restructured its Agent Development Kit samples into google/adk-recipes, which now serves as the single source of ADK content for Agents CLI, Agent Garden, and the Gemini Enterprise app. The repo has three layers: core/ reference patterns that Agents CLI can scaffold, plugins/ with installable vertical skills conforming to the Agent Plugins v1.0.0 spec, and contrib/ with deployable services across five languages. Each recipe includes a validated manifest.yaml and an AGENTS.md written for coding assistants, and contrib and plugin recipes also ship a Dockerfile and a runnability test enforced in CI. Source