Platform Pulse #4: Autonomous Trust and The Agentic Shift

Watch: Identity is the foundation for AI Agents Why identity management is the non-negotiable first step for securing autonomous AI agents in the modern enterprise.

⚡ Serverless & App Modernisation

Scaling beyond the request-response cycle into high-performance background processing.

1. App Engine to Cloud Run Migration: A new preview gcloud beta command simplifies the transition for legacy App Engine apps, automating the containerisation and deployment to Cloud Run.
2. Cloud Run Worker Pools (GA): Now generally available, worker pools provide a dedicated environment for containerised background tasks, reducing the need for complex GKE or GCE management.
3. High-Performance Go on Cloud Run: Technical patterns for squeezing maximum performance out of Go binaries, proving serverless is viable for compute-intensive enterprise workloads.
4. OSOnly24 Runtime GA: The minimalist osonly24 base image is now GA, enabling faster cold starts and smaller attack surfaces for Go and Dart services.
5. 5 Years of Cloud Run Lessons: A veteran architect’s retrospective on statelessness, secret management, and the evolution of the GCP serverless ecosystem.
6. Cloud Shell + VS Code Remote SSH: A guide to using Cloud Shell’s free persistent storage as a remote backend, turning any browser into a full-scale IDE.
7. Artifact Registry Cleanup Policies: Automating image retention and cleanup via Terraform to maintain FinOps hygiene and repository performance.
8. Debugging Cloud Run 403 Errors: A systematic framework for identifying and resolving identity-based 'Forbidden' errors in service-to-service communication.
9. Cloud Build Regional Expansion: CI/CD capabilities have launched in the asia-southeast3 region, allowing for localised builds in Thailand.
10. Workflows in Bangkok: The serverless orchestration service is now live in asia-southeast3, completing the regional stack for event-driven modernisation.

🛡️ IAM & Identity Governance

Rebuilding identity as the 'Operating System' for autonomous AI and ethical trust.

1. Okta: Implementing AI Ethics in Organisations: Moves ethics from theory to practice by establishing cross-functional AI Governance Teams and 'experimentation zones' to foster safe, rapid innovation.
2. Okta: Identity & Authorisation for Autonomous Trust: Defines a new paradigm where every agent action is governed by provenance and attenuation via the Identity Assertion JWT Authorisation Grant (ID-JAG).
3. Okta: Security-Driven Governance: A strategy for unified Identity Threat Protection (ITP) and Governance (OIG) to enable continuous, risk-adjusted access reviews.
4. Cloud KMS Autokey (Public Preview): Automates the creation and management of encryption keys within the same project as the resource, simplifying delegated security.
5. GCP Penetration Testing Guide: A case study on how threat actors chain overprivileged service accounts and legacy keys to exfiltrate data.
6. Service Account Key Cleanup: Leveraging Policy Intelligence to identify and retire unused long-lived keys without impacting production traffic.
7. GTIG AI Threat Intelligence: Recent intelligence on how adversaries use model extraction and AI-enabled social engineering to bypass MFA.
8. UNC1069 Crypto & AI Threats: Analysis of state-sponsored actors using deepfakes and AI agents to breach high-value cryptocurrency platforms.
9. Modern Security: Federated Guardrails: A podcast on the architecture of 'Federated Guardrails'—balancing engineering autonomy with centralised security policy.

🤖 AI Agents & Data Ecosystem

Industrialising intelligence: Bridging data silos with MCP and Conversational Analytics.

1. GEAR Program for AI Agents: Launch of the Gemini Enterprise Agent Ready program, providing specialised training for architects building multi-agent systems.
2. Remote MCP Servers for GCP Databases: Using the Model Context Protocol (MCP) to allow AI agents to interact directly with Cloud SQL, Spanner, and Firestore.
3. Conversational Analytics API: Enables developers to build natural-language interfaces for BigQuery that return both text insights and visual charts.
4. BigQuery Dataset Insights (Preview): Automated generation of relationship graphs and schema summaries to make enterprise data 'agent-readable'.
5. AlloyDB IAM Authentication & Connectors: GA release of language connectors that support managed connection pooling and IAM-integrated security for AlloyDB.
6. Gemini Code Generation at Scale: Key takeaways on leveraging LLMs to generate production-ready code while maintaining security and architectural standards.
7. Mastering Model Fine-Tuning: A deep dive into supervised fine-tuning for Gemini 2.5 Flash on Vertex AI for specialised domain performance.
8. Conductor: Automated AI Code Reviews: A tool for verifying AI-generated implementations against original plans to prevent security and logic bugs.
9. Enterprise AI Platform Architecture: A reference architecture for moving AI systems from PoC to production with a focus on scalability and orchestration.
10. Multimodal RAG Survival Guide: Practical steps for building cost-effective multimodal RAG systems using Gemini and BigQuery vector search.

🌐 Platform Operations & FinOps

Global resilience: Operationalising sovereign cloud and spend-based CUDs.

1. FinOps Guide to Spend-Based CUDs: A strategic guide to maximising cloud savings using the latest commitment-based discount models and validation tools.
2. OTLP support in Cloud Monitoring: Native OpenTelemetry support for standardised, vendor-agnostic observability across hybrid cloud deployments.
3. GDC Air-Gapped Networking (1.15): Advanced IPAM and networking features for strictly controlled, sovereign air-gapped environments.
4. GKE TPU Slice Health Metrics: New system metrics provide granular visibility into the performance and health of TPU clusters for AI training.
5. VMware Engine ve2 Expansion: High-performance ve2 node types are now available in Paris and Osaka, supporting legacy workload modernisation.
6. Flow Analyser Latency Mode: GA release of RTT analysis to help network architects identify and resolve bottlenecks in VPC traffic.
7. Cloud SQL Vertex AI Integration: Simplifies AI application architecture by allowing vector embeddings and predictions directly within Cloud SQL.
8. Log-Based Alerting Strategy: Best practices for configuring proactive alerts that trigger on log patterns to prevent production outages.
9. Service Mesh Security Patches: Critical updates for Cloud Service Mesh images, addressing vulnerabilities in Envoy and Istio.
10. Deepseek on B200 GPUs (GKE): Technical deep dive into high-performance DRANET networking for large-scale LLM inference on GKE.