TRANSPARENT METHOD / SOURCES & LIMITATIONS
How the readiness check works.
An independent, rule-based self-assessment for one workflow. The questions and routing rules are SIGMA’s own; they are not a certified or officially validated assessment instrument.
Method AIR-1.2 · Sources reviewed 2026-10-23
AIR-1.2 uses plain-English questions and everyday examples, including paper, email, Excel and business software. The question meanings, option IDs, prerequisite gates and routing thresholds are unchanged from AIR-1.1. Compatible AIR-1.0/1.1 drafts require explicit resume and keep their original expiry.
What the result means
The check separates a potential workflow opportunity from the people, information and control prerequisites around it. It does not determine readiness across your whole organisation, establish root causes, verify legal compliance or confirm a technical integration. Context is optional and does not set eligibility or infer risk.
There are 21 active questions, or 22 when a declared sensitive or consequential use needs a boundary question. Six groups organise the experience. Allow roughly 6–10 minutes, depending on how much you need to consider; this is an estimate, not a measured completion-time promise.
Six dimensions, with traceable inspiration
The mapping below explains how source concepts informed our original questions and decisions. It does not reproduce an official questionnaire.
| SIGMA dimension | Source concept | Original question focus | SIGMA interpretation |
|---|---|---|---|
| Purpose & opportunity | EDIH strategy; OECD adoption barriers | Goal, frequency and whole-case manual effort | Clarify a recurring opportunity and a baseline; estimates are not ROI. |
| Process & ownership | NIST Map / Govern; EDIH digital readiness | Steps, accountable role, exceptions, task type and workflow-specific evidence | Unclear workflow or expert judgement → process first. No owner → prerequisites. |
| Information & permissions | EDIH data management; NIST risk guidance | Availability, quality and authorised use | No/limited permission, no usable records or poor quality → prepare foundations. |
| Systems & access | OECD digital foundations / secure integration; NIST Govern / Manage | Export or hand-off, proposed output permissions, access/retention/provider controls | Restricted access or missing controls → prerequisites. Unapproved external writes → specialist review. |
| People & capacity | EDIH skills; OECD time/maintenance/skills barriers | Competent reviewer, staff time/funding decision and user involvement | No agreed reviewer or no capacity → foundations; involve users before change. |
| Risk & evaluation | NIST Govern / Map / Measure / Manage; GenAI profile | Declared impact, permitted boundary, whole-workflow measures and fallback | Consequential/serious harm → specialists. Critical unknowns → investigate first. |
Profiles, not an averaged readiness score
No success percentage or overall pass/fail mark is calculated. Each relevant response is classified as a supporting foundation, partial foundation, gap, unknown or descriptive context. Task type is descriptive; choosing AI-related work earns no points. An effort estimate describes the work, not the size of a promised benefit.
- Needs clarification: fewer than half the dimension’s applicable items are informative, rounding the required half up.
- Priority gap: at least half are informative and at least one reports a gap.
- Developing: at least half are informative, none reports a gap and at least one reports a partial foundation.
- Supporting foundation: at least half are informative and the known items support the foundation. This can still include unknowns and is not a safety clearance.
Coverage reports all active questions answered, informative choices and uncertain choices separately. “Not sure” is a valid response and is never substituted for “no”. Unknowns do not earn positive credit. Hidden conditional answers are removed; changing the chosen workflow clears the previous workflow’s responses.
These categories, thresholds and priorities are SIGMA design judgements. They have not been externally validated as a readiness scale.
Route rules—in priority order
- Specialist review needed: declared consequential or serious-harm decisions, unapproved external writes, or sensitive work without an established permitted boundary. Strong foundations elsewhere cannot cancel this gate.
- Prepare the foundations: data use is not fully authorised; owner/reviewer is not agreed; controls, fallback, capacity, usable records or data quality are missing; access is restricted; or a critical prerequisite is unknown. Critical items are permission, owner, reviewer, impact, output permissions, security and fallback, plus the sensitive-use boundary where applicable.
- Prepare the foundations for limited evidence: fewer than three quarters of all active answers are informative, rounding up. This is an uncertainty rule, not a readiness percentage.
- Process first: no single workflow has been selected, steps or exception handling are unclear, minimum inputs are not established, the work depends on expert judgement, task type is unknown, or a recurring opportunity is not yet evident.
- Conventional automation candidate: the prerequisites above are clear and the main work is structured fields or stable rules. Compare simpler deterministic options first.
- AI-assisted workflow candidate: the preceding gates do not apply and the work includes source-based reading, drafting or a mixture of rules and text. This permits considering a scoped assessment only; it is not a pilot or deployment clearance.
The next actions prioritise specialist boundaries and prerequisites. If no workflow is selected, first identify one small task. Missing workflow-specific evidence and inconsistent source records are addressed before general comparison and evaluation steps. Only the first three priorities are shown; the profile and prerequisite details retain other issues. A manual/process improvement or internal preparation may be the appropriate outcome. No route requires purchasing from SIGMA.
Source versions and limitations
NIST AI RMF 1.0 (2023) — Core
Risk guidance: Govern, Map, Measure and Manage are interrelated functions, not four ordered questionnaire stages. NIST states that a revision is in progress.
NIST AI 600-1 — Generative AI Profile (2024)
Risk concepts inform source checking, human review and handling unreliable generated content. This is not a readiness scoring instrument.
OECD SME AI Readiness Tool — pilot (2026)
Conceptual inspiration for digital foundations and barriers. The OECD labels its tool a pilot; its content and results are preliminary. SIGMA does not reuse its questionnaire or scoring.
OECD — Empowering SMEs in the age of AI (2026)
Research context for time, maintenance, skills and secure integration barriers. Its survey covers a non-representative sample of over 2,000 SMEs in 12 countries; it is not a benchmark score for a visitor.
European Commission — EDIH Digital Maturity Assessment (2024)
Concepts include strategy, digital readiness, skills, data and automation. SIGMA uses an independent, narrower workflow check rather than the official EDIH instrument.
NIST’s four functions overlap across risk management; they are not the six groups or a four-step sequence in this check. The OECD pilot’s questionnaire and scoring have not been copied. Source concepts inform this tool; none of these organisations endorses SIGMA or validates its results.
Privacy and reports
There are no required free-text, contact or confidential-data fields. The tool evaluates choices in page memory. It sends no answers to SIGMA, analytics, a model provider or Calendly. The rest of the site still uses existing analytics and normal hosting requests.
Optional saving uses this browser tab’s session storage under a new, versioned key, with a fixed 12-hour expiry from opt-in. It is off by default. Reset or disabling the setting clears that draft. Previous business-health drafts stay separate and are not reinterpreted as these questions. A downloaded report is a local file you control; copying or sharing it is your decision.
Independent preliminary self-assessment informed by OECD/EDIH concepts and NIST risk guidance. Not an official OECD, EDIH, NIST or ISO assessment, endorsement or certification. Self-reported answers are not verified evidence or proof of technical feasibility.