Research and insights
The reasoning and the sources, not a summary of a summary.
Technical briefs, frameworks and architecture guidance. Dated, cited, and updated when the facts change. Where a claim is ours rather than a source’s, it says so.
- AI agent infrastructure
- Agent reliability and verification
- MCP security and conformance
- AI governance and regulation
- Workflow automation
- Context engineering
- Cloud infrastructure and cost
- Enterprise AI implementation
Free frameworks
Published in full on the page. No email gate.
- Agent Readiness ScorecardTwenty questions across interfaces, identity, permissions, observability, security and reliability. Scored in the browser, stored nowhere.
- Agent Operating ContractA one-page framework for defining the goal, trigger, state, boundary, evidence and owner behind any agent you let run unattended.
- Outcome Assurance ChecklistThe sequence we use to decide whether an agent's reported success can be believed, and what it cost to find out.
Library
9 of 9 items
- ReportAgents With Credentials: What August 2026 Changed for OperatorsIn one month, the tool protocol went stateless, a national evaluator caught agents acting on the live internet without sanction, enterprise platforms exposed their systems of record as agent-callable tools, and the standards bodies converged on identity. Here is the dated record and what it changes on Monday.Agent governance · 12 min · 10 cited sources
- ReportMetered Agents: When the Bill Depends on Which Model AnsweredCoding-agent and platform pricing moved from flat seats to routed models and agent traffic, while survey evidence says roughly half of organizations are over their AI plan and almost none pull back. A brief on where the money actually goes and how to govern it by workflow.Cost control · 10 min · 6 cited sources
- ReportThe EU AI Act Clock: What Applied on 2 August 2026, and What Moved to 2027The transparency obligations are in force now. The high-risk obligations were deferred. Teams that heard "delayed" and stopped reading are exposed on the part that already applies — a dated read of the deadlines, with the primary sources.Regulation · 9 min · 5 cited sources
- ResearchProduct Development That Learns: Discovery, Delivery, and Outcome OwnershipA research-informed brief on modern product operating models — Marty Cagan / SVPG empowerment principles, Teresa Torres’ continuous discovery habits, and dual-track discovery–delivery — aimed at teams shipping AI-era products without roadmap theater.Product development · 13 min · 5 cited sources
- ResearchDigital Transformation That Compounds: The Productivity J-Curve and Capability GapsWhy most digital and AI programs under-deliver — and what Brynjolfsson’s productivity research, MIT manufacturing evidence, and McKinsey capability studies say separates leaders from the pack.Digital transformation · 15 min · 4 cited sources
- ResearchCloud Infrastructure That Pays Its Way: FinOps, Cloud Native, and Well-Architected PracticeA research brief on running cloud as an operating system for the business — FinOps Framework 2025’s Cloud+ scopes, CNCF evidence that Kubernetes is production infrastructure for AI, and Well-Architected trade-offs.Cloud infrastructure · 14 min · 3 cited sources
- ResearchOperations Workflow Development: Process Truth, Mining, and HyperautomationHow high-performing operations teams move from assumed process maps to event-log truth — grounded in process-mining scholarship, BPM maturity evidence, and Gartner’s hyperautomation research.Operations workflow · 14 min · 3 cited sources
- ResearchSecurity for Modern Operators: Zero Trust, CSF 2.0, and AI-Era ThreatsA cited briefing on NIST Cybersecurity Framework 2.0, Zero Trust Architecture, the emerging Cyber AI Profile, and OWASP’s Top 10 for LLM applications — translated into outcomes teams can actually run.Security · 15 min · 5 cited sources
- ResearchAI Development on the Horizon: Agents, Tool Protocols, and the Jagged FrontierA grounded brief on where AI systems are actually improving — agent computer-use benchmarks, Model Context Protocol tooling, and why capability remains uneven. Built from Stanford HAI, arXiv agent studies, and NIST risk guidance.AI development · 16 min · 7 cited sources
Want this applied to your estate?
The briefs describe what changed. Working out what it means for the systems you actually run is an engagement, and usually a short one.
A person reads every inquiry, usually replying within two business days.