90-day implementation plan and year-one roadmap →
A phased implementation sequence from initial governance and controls through a year-one programme.
Independent Responsible AI publication
The gap between what agentic AI can do and what we can actually govern.
Controls, evaluation methods, and evidence-graded briefings for the people accountable for AI that acts. Every claim links to its source. Edited by Dr. William Fisher.
Find resources for your work
View the complete directory →Establish governance
Inventory, ownership, risk appetite, impact assessment, and programme planning.
Plan and assess evaluations
Evaluation plans, test validity, red-teaming, and deployment criteria.
Implement technical controls
Permissions, human approval, containment, and multi-agent safeguards.
Monitor and respond
Drift detection, action logs, re-evaluation, and incident response.
Prepare assurance and vendor reviews
Audit evidence, third-party diligence, framework comparisons, and board reporting.
Assess fairness and human impact
Outcome testing, explanations, complaints, and customer remediation.
Implementation resources
All methods →Ten controls to build first →
The compendium’s prioritised starting controls, with links to the full implementation requirements.
Enterprise evaluation programme →
The report’s proposed evaluation programme and the evidence needed to support deployment decisions.
Board-pack checklist →
Evidence and reporting items for a governing-body review of the agentic AI programme.
Research reports
All reports →Agentic Responsible AI Compendium →
112 controls across nine families, with implementation guidance, maturity levels, ownership, and cited sources.
Agentic Measurement & Evaluation: Must-Have Meta-Analysis, 2023–2026 →
An evaluation research review covering test design, benchmark limitations, and a proposed enterprise evaluation programme.
Agentic RAI Controls and Autonomous Workload Safety →
Technical controls for tool permissions, approvals, runtime safeguards, and recovery in autonomous AI workloads.
Enterprise Assurance Stack: What’s New →
Five primary sources considered together as an assurance framework for enterprise governance.
From the editor
A note from Dr. William Fisher
By day I lead the Responsible AI function at a financial-services company: a large, multidisciplinary, international team, and the strategy and R&D behind it. The Observability Layer is not that job. It’s my own work, on my own time, and nothing here speaks for any employer.
I’m a researcher at heart. I built the system behind this site to keep up with a field that moves faster than any one person can read, and then realized the output was useful to more than just me. The problem I’m working in public: how to tell whether an agent is safe to delegate to, and what oversight looks like once it acts. Everything is graded against its sources, so you can check my reading and tell me where I’m wrong.
Dr. William FisherEditorAbout the editor →LinkedIn ↗@DrWilliamFisher ↗
Follow the work
- New briefings by emailEmail
- Dr. William Fisher on LinkedInLinkedIn
- @DrWilliamFisherX
- Daily podcast on Apple PodcastsPodcast
- Daily podcast on SpotifyPodcast
- Every briefing, in your readerRSS
Email requests are handled automatically: you receive each new edition as it is published, and every email carries a one-line unsubscribe reply. Presented by AI hosts; research and editorial direction by Dr. William Fisher.
Browse by subject
All subjects →Agentic systemsEvaluation & assuranceGovernance & accountabilityLaw, policy & standardsSafety & robustnessFairness & human impact
The controls concentrate on agentic AI in financial services. The briefings and research reach across all six subjects.
Recent analysis
Briefing archive →Agent-monitor assurance depends on the adversary tested; NIST comments close October 6
EU Commission lists December 2027 high-risk AI obligations; agent-control evidence demands layered assurance
Agent safety requires tested containment, not merely a monitor; NIST comments due October 6
Govern the Path, Not the Promise
The month’s developments read together, with source references, evidence comparisons, and what remains unsettled.
Source tiers describe authority and rigor, and every finding keeps its limitations attached. Evidence and editorial standards · Edited by Dr. William Fisher.