Google CloudStaying Ahead of Adversarial AI Through Agentic Source Code Review
A practical, agentic, multi-agent code-analysis pipeline that combines LLMs and human-in-the-loop expertise to rapidly model threats and discover vulnerabilities, outpacing adversarial AI in modern software defenses.
Google CloudHow Box is unlocking multimodal enterprise agents with Gemini Embeddings 2
Box teams with Google Cloud to integrate Gemini Multimodal Embeddings 2 into Box's Agentic Platform, delivering layout-aware, cross-format multimodal enterprise agents capable of reasoning over text, images, tables, and charts for richer search and collaboration.
Google CloudBuilding cost-effective, high-throughput gen AI workflows in Google Dataflow
An end-to-end blueprint for cost-efficient, high-throughput gen AI workflows on Google Dataflow, combining a CPU-based pre-filter sentiment classifier with a downstream ADK-backed agent that dynamically orchestrates tools (BigQuery lookups, Gmail API) inside an adaptive Apache Beam DAG to scale, reduce latency, and cut costs.
Google CloudGovernance on autopilot, minus the turbulence
An autopilot governance agent uses column-level lineage to propagate upstream descriptions, glossary terms, policy tags, and trust scores downstream, turning governance from a reactive burden into continuous, self-updating data trust across assets.
Google CloudBuilding operational resilience with agentic AI in financial services
How Deutsche Bank, using agentic AI and a dual-orchestration architecture on Google Cloud, turns manual tabletop resilience into continuous, regulator-ready resilience intelligence aligned with DORA.
Google CloudStaying Ahead of Adversarial AI Through Agentic Source Code Review
An in-depth look at the Agentic Vulnerability Discovery Harness (AVDH): a structured, multi-agent, human-validated pipeline that accelerates proactive code security analysis against adversarial AI and complements continuous CodeMender scanning for a two-layer defense.
Google CloudHow Box is unlocking multimodal enterprise agents with Gemini Embeddings 2
Box enhances its Agentic Platform with Gemini Multimodal Embeddings 2 to power cross-modal, layout-aware enterprise agents that understand text, tables, images, and diverse document formats for richer search and proactive decision support.
Google CloudBuilding cost-effective, high-throughput gen AI workflows in Google Dataflow
Combines a CPU-based pre-filter with a Gemini-powered generative AI agent in Google Dataflow (via the ADK) to deliver cost-effective, high-throughput streaming workflows with adaptive, runtime decision-making.
Google CloudGovernance on autopilot, minus the turbulence
A proactive, lineage-driven approach to data governance that auto-propagates upstream descriptions, business glossary terms, and policy tags downstream to minimize toil and keep data trusted.
CloudflareBGP Role model: tracking the adoption of RFC 9234
Explains how RFC 9234 introduces BGP Roles and the Only to Customer attribute to prevent route leaks, tracks OTC adoption across the Internet, and guides operators to enable RFC 9234 in their networks amid observed OTC-stripping by Tier-1 providers.
CloudflareBGP Role model: tracking the adoption of RFC 9234
Tracking RFC 9234's BGP Role and OTC adoption worldwide, this post analyzes how OTC is deployed, where it's being stripped, and the implications for preventing route leaks in real networks.
Apple MLA Specialized Semismooth Newton Method for Kernel-Based Optimal Transport
A specialized semismooth Newton method for kernel-based optimal transport using a nonsmooth fixed-point formulation to achieve global convergence and substantial speedups over SSIPM.