Agentic AI Safety Research with Agent Bulkhead
We are providing resources and practical expertise to support Agent Bulkhead's work in agentic safety and AI research, bringing a production perspective to the design of safer AI systems.
Read articleARTICLES
Notes, methods, and field reports from Graylark Labs.
Longer-form writing from active projects across Graylark LRM, InsightMesh, Graylark Polyglot, EU AI Act classification, and enterprise-safe autonomous execution supporting Graylark Technologies.
Reading the EU AI Act as an Engineering Specification for Workplace AI
Deployer obligations for high-risk workplace AI translated into system requirements: provenance-grade logging, human oversight as UX, worker notification workflows, and change management with a loop in it.
Read articlePay Equity as a Data Engineering Problem: Preparing for Directive (EU) 2023/970
The Pay Transparency Directive as a data brief: equal-value categories as data modelling, pay assembly across fragmented systems, small-group discipline, and reproducible runs that survive scrutiny.
Read articleCompliance as Code: How We Built Our Own Automated ISMS and Pen-Testing Framework
Why we built our own automated compliance-evidence and penetration-testing framework instead of buying a platform - and why continuous, evidence-based proof can stand alongside a once-a-year external pen test.
Read articleHow We Built Model Forge: Autonomous Training Infrastructure for Domain-Specific LLMs
How we built autonomous training infrastructure using autoresearch, knowledge distillation, and structured legal training data to produce domain-specific LLMs for European labour relations.
Read articleBuilding a Production-Grade MCP Server Inside a Multi-Tenant Enterprise Platform
How we integrated MCP into Graylark LRM with tenant-safe access, dual-mode authentication, and production-grade audit and governance controls.
Read articleInside InsightMesh: Building Country-Specific Labour Intelligence That Stays Grounded
How we connect change context, historic outcomes, legal advisories, and proprietary models into grounded country-level intelligence.
Read articleGraylark LRM: A Labour Relations Platform Built for Complexity, Compliance, and AI
Why we built Graylark LRM, what is technically novel, and how we deliver controlled AI across complex labour-relations workflows.
Read articleGraylark Polyglot: Building Domain-Grounded Translation for Labour Relations at Scale
Why multilingual enterprise delivery needs domain context, and how Polyglot reduces translation time and cost for Graylark LRM.
Read articleTraining Graylarks EU AI Act Classifier: Fast, Accurate Insights at Scale
How we approach model training, fine-tuning, uncertainty control, and low-latency inference for high-volume governance workflows.
Read article