For CIOs, CTOs and enterprise change leaders.Fictional data · runs in your browser
Break glass fast. Leave a trail your auditor can follow.
Know what a major change touches before you approve it. AI prepares the record; a named human decides.
Open the demoOne fictional decision, seven stages.
Decision record · CDR-DEMO-0001Fictional
Deploy Microsoft 365 Copilot enterprise-wide
ScopeAsked: 12,400 users. Recommended: ~4.8% of seats.
Evidence7 items1 missing
DissentCISO, CFO, Legal and Red Team, kept verbatim.
Conditions6 owned, 0 verified
AI authorityRecommend only
Decision ownerAwaiting approver
People decide. AI recommends.Gaps shown, not smoothed over.
01The executive problem
Six forums. Six slices. No one decision.
01Shadow changes leave the map out of date. Fixes made outside the process mean the CMDB describes a system that no longer exists, so the next decision is made on a picture that's already wrong.
02The knowledge sits in people's heads. When a critical system fails, the decision waits on the one engineer who knows how it really fits together, or defers to them without checking.
03Approvals turn into rubber stamps. When every change gets the same scrutiny, reviewers stop scrutinising. The risky change gets waved through with the rest.
04Emergency fixes leave no trail. Break-glass changes bypass the normal controls. If no one brings them back for review, auditors find a gap where the evidence should be.
CABSecurityRiskLegalFinanceArchitecture
One decisionToday: partial viewsboardlevel: one evidenced decision
02boardlevel and the CAB
Your change authority decides. BoardLevel prepares.
BoardLevel doesn't replace your CAB or eCAB. It's designed to close the gaps between them. Policy decides which changes need people at all. The impact map distinguishes what's known, observed, supplied, inferred, stale or missing before anyone approves. And every break-glass change is brought back into a versioned record, reconciled and reviewed.
Evidence BoardLevel can reason across
CMDBArchitectureCloudIdentityNetworkHRFacilitiesRisk and legalSuppliersProjects
boardlevel preparesImpact · unknowns · dissent
Your CAB + named executiveDecide and own it
03How it works
One decision, seven stages, one record.
A record auditors can followEvery approval is tied to a specific proposal version, with the evidence, dissent and conditions attached. Emergency changes get a return path: what was done, who authorised it, and when it was reconciled and reviewed.
Policy sets the bar, not the meetingEach change type carries a defined ceiling for AI and a required set of approvers. Low-risk changes keep moving, and people spend their attention on the changes that can actually hurt you.
Knowledge out of heads and into the recordThe impact map marks each dependency as observed, supplied or inferred, and names who holds what isn't written down. A 2am decision shouldn't depend on who answers the phone.
01ProposalChange, scope, benefit
02Impact mapEverything it touches
03Evidence trailEvery claim sourced
04Specialist panelIndependent challenge
05Open challengesDissent kept verbatim
06Executive reviewOne view, not six
01 ProposalChange, scope, benefit
02 Impact mapEverything it touches
03 Evidence trailEvery claim sourced
04 Specialist panelIndependent challenge
05 Open challengesDissent kept verbatim
06 Executive reviewOne view, not six
See the 10 impact layers+
Strategy and capabilitiesProcessesPeople and rolesUnknownApplicationsIntegrations and dataCloud and on-prem infrastructureNetworksUnknownSecurity controls and trust boundariesSuppliersSites and buildingsUnknown
04Human control
People decide. AI recommends.
For major change, AI stops at Recommend. No one holds two roles.
A1Assist
A2 · CeilingRecommend
A3Execute after approval
A4Bounded delegation
A5Adaptive under policy
AI proposes with rationale. A human decides and acts.A3–A5 not in pilot scope
See the six roles+
RoleDoesCannot
ProposerDoesSubmits the changeCannotAssess or approve
ChallengerDoesTests it, records dissentCannotEdit the proposal
Risk ReviewDoesSets residual riskCannotApprove
ApproverDoesMakes the named decisionCannotPropose or execute
ExecutorDoesDelivers approved scopeCannotChange scope
AuditorDoesVerifies the recordCannotAlter it
05Security and data
What runs where, and who decides.
This prototypeRuns in your browser. Fonts load from an external font service.
Who decidesA named human, tied to a proposal version.
Pilot dataDocuments and context you choose to share for the selected decision; no production-system connections required.
06Design partners
Bring us one material or emergency change.
We'll reconstruct the decision and show you what your existing process could see, what was missing, and what sat between the gaps.
Stays on your device
Pilot termsDesign partners
Length6 weeks
Your effortSponsor ~2 hrs/week · 3–5 specialists, ~1 hr each
CostFree for selected partners
You getOne live decision record + debrief