Consultant workbench · guided demo
How a readiness assessment and an internal audit run in Assure, the workbench behind ENFTC's AI governance work — from the first client conversation to a corrective action verified on fresh evidence.
Client: Northstar Services, a fictional demo client. Every name and record shown is invented.
Guided demo
Follow one requirement — approval before an AI workflow goes live — through a readiness assessment, then an internal audit of the same client. Pick any stage, or let it play.
Readiness assessment
Internal audit · separate engagement
Remediation
Live workbench
Stage 01 / 09 · Readiness assessment
A guided interview captures the organisation, jurisdictions and the boundary of the work. The summary builds as we talk, so nothing is assumed.
Human decision
The scope is agreed with you and saved explicitly. It stays a working scope until we refine it.
Stage 02 / 09 · Readiness assessment
Each system gets an intended purpose, an owner, a lifecycle stage and a working role under the EU AI Act. The inventory feeds every later step.
Human decision
Roles stay provisional until reviewed. The workbench never assigns a statutory risk category.
Stage 03 / 09 · Readiness assessment
32 ISO/IEC 42001 requirements and a per-system EU AI Act analysis, each tied to its page reference and a working question.
Human decision
Each item needs an explicit completed review. A count is recorded work, never a compliance percentage.
Stage 04 / 09 · Readiness assessment
Documents are read in place on the workbench’s own server. The judgement, its rationale and its limits are saved against the exact file version.
Human decision
Gap identified, with a written rationale, by a named reviewer. A follow-up action is created from it.
Stage 05 / 09 · Readiness assessment
A finding keeps its observation, evidence basis, owner, due date and action plan together until it is closed.
Human decision
Priority is work priority, not an audit grade. Closing a finding needs closure evidence and a rationale.
Stage 06 / 09 · Readiness assessment
The executive summary, priorities and limitations are prepared from the saved records and kept exactly as reviewed.
Human decision
Released by a named reviewer. The limits of the evidence are stated, not hidden.
Stage 07 / 09 · Internal audit
For clients whose AI management system is in place, an internal audit is scoped separately. A reviewed audit plan drives recorded sessions and observations against the stated criteria, on the same evidence base.
Human decision
Observations are recorded by the auditor. Labels organise work; they do not create findings.
Stage 08 / 09 · Internal audit
Root cause, correction, implementation and fresh samples lead to an explicit effectiveness decision that closes the case.
Human decision
Effective on three new samples and one blocked test, and no further than that.
Stage 09 / 09 · Remediation
The drafting assistant proposes wording for the client’s own procedure, one section at a time, and the coverage check shows which requirement each section cites and which references still need checking.
Human decision
I accept, edit or reject each proposal. I lock the reviewed version, and the client’s sign-off is recorded against it.
Closing note · the live workbench
The live workbench is open only to people I invite, and every sign-in asks for a password and a code from an authenticator app.
Human decision
There is no self sign-up. An account exists only when I send an invitation.
For collaborators
The live workbench is for invited users and testers. Each person signs in with a password and a code from an authenticator app.
Not invited yet? Ask by email and I’ll set up your access.
Describe the AI system, the decision you need to make and any deadline. You'll get a proposed scope, the input needed and a fee before you commit.