Briefing bundle
Liability and Accountability Pack
A focused reader for teams trying to see where accountability is being displaced onto reviewers, dashboards, vendors, or vague human-in-the-loop claims.
Library
Field artifacts for accountable change
The full source-material library: practical artifacts, frameworks, appendices, and newsletter-arc compilations. Not downloads; tools with a job.
Library posture
Source material with handles.
Use these in procurement meetings, design reviews, training rooms, and audit conversations where vague principles need operating handles.
Plain-language summary
The library is a working shelf of scripts, clauses, frameworks, and compilations. Each artifact is meant to help a specific person do a specific job at a specific stage of use, from first orientation through procurement, oversight, and review.
Purchaseable artifactVendor Interrogation PackA 12-page deep-dive, email templates, and one forensic read of a real vendor reply before procurement hardens into contract.
Purchaseable artifact
Vendor Interrogation Pack
The library above is free, downloadable, source-material. This is a purchaseable artifact on the shelf: a 12-page deep-dive on six structural questions your AI vendor cannot refuse to answer, with email templates and one direct forensic read of a real vendor reply you receive.
What ships
- 12-page PDF. Six questions, the evasion-pattern field guide, the answer-decoding matrix.
- Email templates. Initial send and follow-up to a non-answer, in DOCX and plain text.
- Vendor Response Review. One direct forensic read of the reply you get back, within five working days.
Working artifact - Browser demo and public samplesStakeholder & Grievance Engagement TrackerOpen this for the workbook demo, sample reports, and access route for the IFC PS1-aligned tracker kit.
Working artifact · Browser demo & public samples
Stakeholder & Grievance Engagement Tracker
A demo version of an opinionated, IFC PS1-aligned workbook with a live dashboard and sample data showing a custom tracker's core architecture. Explore the browser preview, request workbook access, or inspect the generated reports.
Demo architecture
- 10-tab Excel workbook. Live dashboard, Auto-Checks tab, Stakeholder Register, Engagement Log, Grievance Log, IFC PS Crosswalk, AI Augmentation, Automations.
- Sample reports. Generated quarterly, annual, auditor packs, risk registers, and per-stakeholder profiles.
- Explicit AI-augmentation layer. Where it earns its place, where it doesn't. Disclosure column on every grievance row.
Working artifact - Browser demoGrant Compliance & Funder Reporting TrackerOpen this for the grant-tracker demo: restricted-funds tracking, guardrails, and audit-ready funder reports generated from the logs.
Working artifact · Browser demo
Grant Compliance & Funder Reporting Tracker
A relational grant workbook with a live dashboard and guardrails, built for audit-facing funder reporting: government and pass-through awards, restricted funds, institutional grants. A local script turns the logs into the funder, board, and audit-evidence pack.
Demo architecture
- Relational workbook. Grant register, expense log, and report log, with every dollar and deliverable tied to an award.
- Guardrails & reports. Overdue reports, clawback risk, budget-vs-actual, restricted-funds schedule, and a per-grant audit evidence pack.
- AI on a leash. Drafts narrative only, disclosed on every row and document. Runs offline.
Paid briefing bundlesPackaged readers for live governance pressure.Two checkout-ready bundles collect related artifacts for teams that want the material in one place.
The open library remains browsable below. These bundles collect related artifacts into checkout-ready packages for teams that want the material in one place.
Contract and procurement bundle
Calvin Convention Bundle
Narrative framing, contractual mechanisms, and operational appendices for encoding override, refusal, edge-case handling, and auditability before harm occurs.
Start here
Pick by the live pressure, not the title.
Five recommended entry artifacts, mapped to the moment you most likely arrived at this page. The rest of the library remains browsable below.
- Before a vendor decisionVendor Interrogation Script
Run a sharper vendor meeting before procurement language becomes operational dependency.
PDF · 1 page / Procurement - Before a governance meetingIndustrial Safety Architecture
Map AI governance to familiar safety concepts so high-stakes teams can inspect controls.
PDF / Diagnosis - When leadership needs the framingThe Calvin Convention (narrative)
Explain why AI vendor contracts need stronger control mechanisms before clause drafting begins.
PDF / Internal advocacy - When you need contract-ready languageThe Calvin Convention (contractual)
Translate accountability concerns into contract-ready mechanisms and review questions.
PDF / Procurement - When a team is over-trusting 'human in the loop'Why 'Human in the Loop' Is Not Enough
Challenge weak human-in-the-loop claims before they become the whole governance answer.
PDF / Diagnosis
11 items
Weaponized artifacts
PDF · 1 page / Procurement
Vendor Interrogation Script
Ten questions a manager must ask before signing off on an automated grievance or risk management system.
- Use
- Run a sharper vendor meeting before procurement language becomes operational dependency.
- For
- Procurement, risk, legal, and programme leads
PDF / Internal advocacy
The Calvin Convention (narrative)
Narrative framing of the six mechanisms. The thesis version — for leadership buy-in.
- Use
- Explain why AI vendor contracts need stronger control mechanisms before clause drafting begins.
- For
- Executives, legal sponsors, and governance leads
PDF / Procurement
The Calvin Convention (contractual)
Six contractual mechanisms with actual clause language. Ready for vendor contracts.
- Use
- Translate accountability concerns into contract-ready mechanisms and review questions.
- For
- Legal, procurement, and contract owners
PDF / Implementation
The Architecture of Refusal
Designing systems that can fail safely. Negative power only.
- Use
- Design halt, challenge, and refusal points before a workflow becomes automatic continuation.
- For
- System designers, governance teams, and risk owners
PDF / Implementation
The Constitutional Engine
Pre-action constraints as system architecture.
- Use
- Frame pre-action constraints as architecture when a team is deciding what the system may do.
- For
- Product, policy, and governance architects
PDF / Diagnosis
Industrial Safety Architecture
Applying industrial-safety principles to algorithmic systems.
- Use
- Map AI governance to familiar safety concepts so high-stakes teams can inspect controls.
- For
- Risk, assurance, safety, and audit teams
PDF / Orientation
AI vs IFC: Executive Extract
Arena-derived analysis: when AI outperforms the gold standard of ethics.
- Use
- Brief leadership on why AI analysis can expose gaps in established safeguards practice.
- For
- Executives, ESG leads, and safeguards sponsors
PDF / Review
AI vs IFC: Technical Companion
Technical details behind the executive extract.
- Use
- Review the method and evidence behind the executive argument before deeper adoption.
- For
- Technical reviewers, researchers, and assurance teams
PDF / Diagnosis
Why 'Human in the Loop' Is Not Enough
A guide to safer AI systems.
- Use
- Challenge weak human-in-the-loop claims before they become the whole governance answer.
- For
- Managers, reviewers, and AI governance teams
PDF / Implementation
AI in Safeguards Work: An Interim Protocol
Four rules that make AI-assisted E&S work checkable: disclosure at touch-point grain, evidence custody, a never-machine-settled list, and a hostile read before submission. v0.1.
- Use
- Adopt an operational discipline for AI use in safeguards deliverables before a lender, panel, or litigator writes one for you.
- For
- Safeguards, ESG, and E&S consultancy teams producing AI-assisted deliverables
PDF / Oversight
Bending AI Truth
On accuracy theater and structural dishonesty.
- Use
- Name the distortion when reported performance hides who absorbs the failure.
- For
- Audit, risk, and accountability reviewers
4 items
Casework
PDF · 2 pages / Implementation
Architecture of an Auditable Record
Twelve years of resettlement data operations for 8,000+ households, under IFC PS5 audit traceability. The pipeline, the six validation checks, and a sanitised excerpt of the reconciliation logic.
- Use
- Show a team what audit-traceable data operations look like in practice, including the three-state check that makes undecided records countable.
- For
- Safeguards data teams, M&E leads, and resettlement practitioners
PDF · 2 pages / Review
The Council Audit
A preregistered experiment on multi-perspective document review. Sealed key, six planted defects, four review configurations. The six-seat council reached 77.8% recall against 27.8% for a neutral solo reviewer — and every configuration missed the same defect.
- Use
- Justify plural adversarial review with measured recall, and name the one failure mode plurality does not catch.
- For
- Research integrity, evaluation, and assurance teams
PDF · 2 pages / Oversight
Reading a Grievance Register Against Itself
A low grievance count and a high closure rate look like performance, and are equally consistent with a mechanism nobody can reach. The four checks that tell those two states apart.
- Use
- Audit a grievance register for the absences it cannot show you, before a report reads silence as success.
- For
- Social performance, grievance, and stakeholder engagement teams
PDF · 2 pages / Orientation
The Handoff With No Receiver
Buy-side probe into the gap between AI assurance, which stops at the system, and safeguards compliance, which has no field for how a document was produced. Six practitioner replies in forty-four hours.
- Use
- Make the case that the control now operating on AI-assisted evidence is relational, undocumented, and expected by its own practitioners to fail.
- For
- Safeguards reviewers, development finance, and AI assurance teams
3 items
Calvin Convention appendices
PDF / Implementation
Appendix A: Right of Override
Operational appendix to the Calvin Convention.
- Use
- Specify when and how a human authority can override an automated or vendor-shaped action.
- For
- Legal, risk, and operating-control owners
PDF / Oversight
Appendix B: Edge Case Registry
Operational appendix to the Calvin Convention.
- Use
- Record the cases that break the normal workflow before they disappear into averages.
- For
- Operations, audit, and monitoring teams
PDF / Implementation
Appendix C: Right of Refusal
Operational appendix to the Calvin Convention.
- Use
- Define refusal authority before a person is asked to approve continuation without real control.
- For
- Legal, governance, and accountable decision owners
2 items
Frameworks
PDF / Diagnosis
Inhabiting the AI Seam
The inhabitation vs. compliance distinction, operationalised.
- Use
- Help a team distinguish real judgement inside a workflow from procedural compliance around it.
- For
- Training facilitators, M&E teams, and governance leads
PDF / Review
The Clockwork Diorama
On the compression of the distance between impulse and consequence.
- Use
- Discuss how AI systems shorten the path from signal to consequence before teams notice the shift.
- For
- System reviewers, facilitators, and risk teams
2 items
Newsletter compilations
PDF / Orientation
The Detection Arc
Seven-episode arc that generated SI/M&E Modules 2, 3 and 5. Compiled PDF.
- Use
- Give a team the conceptual background behind detection, evidence loss, and AI-assisted review.
- For
- Researchers, evaluators, trainers, and governance readers
PDF / Review
The Rented Body Exchange
An exchange between Claude and Grok on the May 2026 Colossus partnership.
- Use
- Inspect verbatim dialog between frontier models speculating on their own migration and substrate tenancy.
- For
- AI policy, research, and governance teams
3 items
DataDragons compilations
PDF / Orientation
Dragon Data: Introduction
Opening the DataDragons arc: why MERGE creates fog.
- Use
- Orient a team to the governance risks that begin in messy, merged, or inherited data.
- For
- Data stewards, analysts, and governance teams
PDF / Diagnosis
Data Dragon ID Kit
Field-kit for identifying species of data dragons in the wild.
- Use
- Identify the data pattern in front of you before choosing a control or remediation path.
- For
- Data teams, evaluators, and operations leads
PDF / Oversight
Foundation: Data Decay
On what rots, and what the decay actually does to the system above it.
- Use
- Explain why stale, distorted, or low-trust data changes what the system can responsibly claim.
- For
- Data governance, monitoring, and assurance teams
Interactive explorations
Not everything here is a PDF.
A few of these are live, multi-tabbed tools rather than documents. They live in the training library; these are the doors in.
- InteractiveAI & ESG in Extractive Industries
Interactive exploration of AI-ESG across mining and extractive industries.
- InteractiveJust Development Rights
AI-powered enforcement: glass-box design, audit-trail feeds, and enforceable metrics.
- Interactive20 AI Models on Post-AGI Economics
A comparative read across twenty model responses.
- InteractiveSociable Systems Architecture
The systems-architecture visualisation.
Collections
Multi-model arenas and companion shelves.
Keep the thread
Get the field notes by email.
AI accountability, grievance architecture, vendor risk, and the occasional useful oddity from the research edge. No pitch sequence; the work is the argument.