Twenty-seven architectural questions that distinguish leading organizations from trailing ones in 2026. Score your current state, set your target state, and see the gap on a single chart. Not the questions you've already answered, the questions most leadership teams don't yet know to ask.
Most agentic readiness conversations skip past the architectural questions because executives don't know they need to ask them. Do you have a strategy? — yes. Do you have governance? — yes. Do you have executive sponsorship? — yes. Those answers used to be diagnostic; in 2026 they're table stakes. The questions worth asking now are about the depth of architectural commitment behind the strategy: vendor neutrality as a property of your stack, what data your agents send to frontier models, what controls prevent runaway costs, how fast you can launch a new agent with full memory, what gates protect your AR system from a wrong invoice. This assessment focuses on those questions. It also explains why each one matters.
The dimensions on the chart all assume a baseline of strategic commitment. If your organization hasn't yet established that baseline, the architectural questions further down won't help you — they'll just produce a low-score chart that doesn't tell you what's actually wrong. Three preamble questions establish whether you're ready for the diagnostic, before you start.
Proceed to the assessment. The architectural diagnostics will be the right next conversation.
Pause. The architectural diagnostics won't help yet. The strategic prerequisites need to be established first; the assessment will produce a chart that tells you nothing you don't already know. Bookmark this page and come back when those preamble answers change.
Why these three and not others. Vision, culture, executive engagement, ethics committees — these all matter, but they don't pass the diagnostic threshold. Self-reported answers cluster too tightly to distinguish meaningful differences. The three preamble questions are the ones where a "no" actually changes what an organization should do next. Everything else either has consensus answers (and adds noise to the chart) or doesn't have a self-report-friendly signal.
Each goal addresses a distinct layer of agentic-enterprise architecture where measurable variance still exists in 2026. The six goals don't cover every conceivable dimension — they cover the dimensions where the answer matters and where the variance is real.
The dimensions cluster around two kinds of value. Some are dimensions where organizations are visibly diverging today — most organizations don't yet do pre-deployment evaluation, but some do, and the gap is large. Others are dimensions where almost no one yet has a strong answer — almost no organization has thought seriously about audit format portability, agent launch velocity with inherited memory, federation across sovereign boundaries, or cost controls at the inference boundary. Both kinds matter.
Governance is not one setting you switch on. Autonomy, human oversight, the metric that matters, and the gate you have to clear all shift as an agent moves from a first experiment to enterprise scale. This is what changes at each stage, and where Operon holds the line.
More autonomy. Higher stakes. One governance layer at every stage.
Operon holds the governance gate at each stage, so autonomy and oversight advance together instead of trading off. The same policy model, decision trace, and human authority carry from the first experiment through board-level assurance.
Twenty-seven spokes, ordered by gap. The solid blue stem is the maturity you have; the ember track continuing past it is what is missing, out to the target ring. Widest gap sits at twelve o’clock. Full names and goal clusters are in the table below.
Each of the 27 dimensions is scored on a 0–5 ordinal scale with half-point intermediates available, producing 11 possible scores per dimension. The half-point intermediates let respondents express "between proactive and strategic" without forcing a binary choice.
| Score | Level | Plain English |
|---|---|---|
| 0 | Non-existent | We don't do this; haven't considered it. |
| 1 | Reactive | We do this when forced to, after problems occur. |
| 2 | Emerging | We're starting to formalize this; some early practice exists. |
| 3 | Active | We do this consistently; it's documented, owned. |
| 4 | Proactive | We do this systematically; it's measured, improving. |
| 5 | Strategic | We do this as a competitive advantage; it's a stated business priority. |
Two scores per dimension, not one. The respondent records current state (where the organization actually is today) and target state (where the organization intends to be at a defined milestone — typically 12 to 24 months out). The gap between current and target is the actionable artifact. A 2.5 → 4.5 gap on Data Egress Controls is meaningfully different from a 4.0 → 4.5 gap on the same dimension; the spider chart shows both.
Hierarchical aggregation. Multiple respondents from the same organization can take the assessment. Individual scores aggregate to role-level (engineering, security, compliance, business, finance), which aggregate to operating-company scores, which aggregate to enterprise scores. Where role perspectives diverge meaningfully — typically the security team scoring Runtime Governance differently than the product team, or finance scoring Operational Economics differently than engineering — the divergence is itself diagnostic.
Methodology note. The dual current/target scoring, hierarchical aggregation, and trend tracking patterns are adapted from US Patent 10,997,532 ("System and Method for Assessing and Optimizing Master Data Maturity," 2018), originally developed for master data maturity assessment. The dimension model is new; the assessment architecture is proven.
The wheel ranks dimensions against each other. This answers the other question: which goal cluster is carrying the weight. Shading is cut across your own range rather than a fixed scale, so the darkest cells are your worst — not everyone’s. Every cell prints its number regardless.
Sorted by gap size. The dimensions with the biggest distance between current and target are usually where investment lands.
| Goal | Dimension | Current | Target | Gap |
|---|
The results above are the artifact of the assessment. A respondent who completed 27 dimensions earned something they can use — not a screen they have to screenshot. Three things appear immediately when the assessment is done, in this order, on this page.
The 27-spoke gap wheel, ranked widest gap first, with the cluster field beneath it. Below those, the live gap table, sorted by gap size descending. The line a respondent can repeat to their boss in 30 seconds: "Our three largest gaps are X, Y, and Z."
No email gate. No signup wall. The PDF downloads on click.
All 27 dimensions in the gap table above, sorted by gap size descending. Each row links to the “What most executives often miss” expandable on the dimension card. Each gap also shows which ROI category it maps to — a visual reinforcement of the connection between maturity gap and dollar value.
Your assessment reflects one perspective. Get scores from your security, compliance, and finance teams to see where role perspectives diverge. Included for Operon customers.
Set up organizational modeArchitecture review with our team. We'll walk through how Operon's platform addresses each of your largest gaps and what implementation would actually look like in your environment.
Book an architecture reviewBookmark a private link to come back to this assessment, or save it to track maturity trajectory over time when organizational mode goes live.
The PDF is the artifact a respondent forwards to their CFO, board, peer leader, or saves to revisit in six months. It's designed to be readable on its own — someone who didn't take the assessment should be able to read the PDF and understand the diagnostic. Approximately 8–10 pages, generated client-side, no data leaves the browser.
Organization name (or "Anonymous"), assessment date, respondent role, completion timestamp. Operon Studio attribution in small type. Light on branding.
The single most important page. Average current/target scores, the three largest gaps named explicitly, what each gap means, and the ROI handoff line.
Full-page chart with current and target state overlaid. Goal-cluster colors annotated. Legend at the bottom.
Six rows, one per goal: average current, average target, gap, and a horizontal-bar visual sorted by gap size descending.
All 27 dimensions, grouped by goal cluster. Each entry shows scores, level-0/5 endpoints, the “most executives often miss” callout in full, and the ROI category mapping. The meat of the document.
Scoring model (0–5 ordinal, half-point intermediates, dual current/target). Reference to US Patent 10,997,532. Disclaimer that this is an educational diagnostic, not an audit.
Three options with URLs: run the ROI calculator (pre-populated link), take organizationally, schedule architecture review.
The single page where Operon Studio's voice appears explicitly. One paragraph, not pushy. Why this assessment exists, what the landscape of existing models looks like, and the gap this one addresses.
Visual style. Clean, professional, readable. Same colour language as the gap wheel. Plenty of whitespace. Tables and charts, not walls of text. The PDF should feel like a McKinsey diagnostic deliverable, not a marketing brochure. Filename pattern: Operon-Maturity-Assessment-[Org]-[Date].pdf.
Each of the six goals maps to one or more ROI calculator categories. The maturity gap on each goal predicts the size of the savings opportunity. Lower current maturity means higher gross burden today, which means larger Operon savings when the gap closes.
| Maturity Goal | Maps to ROI Category | How the gap drives the calculation |
|---|---|---|
| Agent Lifecycle Discipline | Effort to Maintain Agents | Lower current maturity = higher gross maintenance burden today = larger savings opportunity from versioning, rollback, and decommissioning discipline. |
| Runtime Governance | Policy Violation Avoidance + Action-related incidents | Lower current maturity = higher count of avoidable violations and runaway actions today = larger savings opportunity from runtime gates, action validation, and scoped authorization. |
| Cross-Platform Coverage | Shadow-AI Incidents + Platform Consolidation | Low current maturity unlocks the shadow-AI category in the calculator. Builder Sprawl maturity drives the platform-consolidation savings line. |
| Data Sovereignty & Egress | Compliance exposure + Audit efficiency | Low current maturity on Egress Controls drives compliance incident exposure; low maturity on Schema Openness and Decision Provenance drives audit efficiency savings. |
| Evidence & Audit | Compliance Audit Efficiency | Lower current maturity = higher gross audit-evidence-assembly cost today = larger savings opportunity from EDT-backed automated packs. |
| Operational Economics | Effort to Maintain Agents + Productivity + Cost-control savings | Cost Observability and Controls maturity directly affect frontier-model spend management. Agent Launch Velocity affects productivity. Knowledge Persistence affects per-agent build cost. Drift Response affects maintenance leverage. |
The integration in practice. When a respondent completes this assessment, the calculator on /roi-calculator can be pre-populated with defaults reflecting their current-state scores. A respondent scoring 1.5 on Cost Observability with a target of 4.5 sees calculator defaults reflecting "high current frontier-model spend, no per-agent attribution, large gap to close" — typically meaning higher gross inference burden, more candidate avoidable spend, and a larger consolidation opportunity. The calculator output reflects the size of the gap, not a one-size-fits-all default.
This is the mechanism that turns the calculator from a number-generator into a tailored business case. Two organizations with identical agent counts but different maturity gaps see meaningfully different first-year value numbers, and both numbers are defensible because the inputs trace back to a specific assessment.
This assessment is not a replacement for the existing maturity frameworks. Most organizations evaluating agentic readiness will encounter several of them; understanding what each covers helps you use them complementarily rather than choosing one and discarding the rest.
| Framework | What it covers well | What it structurally cannot ask |
|---|---|---|
| Microsoft Agentic AI Adoption Maturity Model (Copilot Studio guidance) | Five-level CMM-based progression for organizations adopting Microsoft's Copilot ecosystem. Strong on operational readiness for that platform. | Cross-vendor coverage, vendor neutrality, data egress to non-Microsoft model providers. The model assumes Microsoft as the platform; it cannot meaningfully ask about agents running outside that perimeter. |
| Salesforce Agentic Maturity Model | Four-stage roadmap aligned to Agentforce and Atlas adoption. Useful for Salesforce-committed shops scaling into agent workflows. | Vendor neutrality, builder sprawl across non-Salesforce frameworks, federation across operating boundaries that Salesforce doesn't span. |
| Gartner AI Maturity Model | Seven-dimension analyst framework: strategy, product, governance, engineering, data, operating models, culture. Vendor-neutral; broadly respected. | Operational specificity. By design, the framework is high-level. The architectural questions about format portability, federation posture, drift response, decision provenance, data egress controls, action validation gates are not at its level of granularity. |
| MIT CISR Enterprise AI Maturity Model | 0–100% scale academic research on cumulative enterprise AI capability building. Strong empirical grounding. | Agent-specific architectural questions. The MIT CISR model is enterprise AI generally; many of the dimensions in this assessment are below its level of resolution. |
| AAGMM (Acharya, 2026) | Five-level governance maturity model spanning 12 governance domains, grounded in NIST AI RMF and ISO/IEC 42001, validated through 750 simulation runs. Most architecturally serious of the published frameworks. | Coverage for cross-platform federation, format portability, vendor neutrality as architectural property, semantic-vs-physical binding for drift response, frontier-model data egress, cost observability. AAGMM focuses on governance domains; this assessment focuses on architectural properties that determine whether governance dimensions are achievable. |
Dimensions where the answer matters and the variance is real. Some dimensions overlap with existing frameworks (Pre-deployment Evaluation, Policy Enforcement, Identity, Audit Trace Completeness). Others are dimensions other models structurally cannot ask: Builder Sprawl, Vendor Neutrality as an architectural property, Schema Openness, Federation Posture, Data Egress Awareness and Controls, Action Validation Gates, Cost Observability and Controls, Agent Launch Velocity, Knowledge Persistence with governance.
Establish a five-level CMM-style progression with prescribed practice-by-stage maps. The CMM-progression assessment is what Microsoft and AAGMM do well; this one focuses on diagnostic resolution rather than progression structure. Use both.
The same architectural pattern from the patent applies here: assessments are stored over time, and the trajectory of the organization's maturity is tracked using a least-squares fit on dimension scores. A maturity trend score for each dimension shows whether the organization is improving, holding steady, or regressing.
Most organizations adopting agentic AI are moving through maturity, not sitting at it. A score of 2.5 on Runtime Governance is a different signal depending on whether the organization was 1.0 a year ago (improving fast) or 3.5 a year ago (regressing — typically because someone left or a reorganization disrupted the discipline). The trend score surfaces direction, not just position.
The trajectory view shows current state, six months ago, twelve months ago, and the projected trajectory based on rate of change. This becomes valuable for board reporting, internal investment cases, and external compliance disclosure — some EU AI Act provisions reward demonstrable trajectory of governance maturity over time, not just point-in-time posture.
The assessment can be taken individually or organizationally. Both modes use the same 27-dimension framework.
The assessment takes 25–30 minutes and produces a real artifact — a ranked gap wheel and a 10-page PDF — that you can take into your own internal conversation without us. The architectural questions surfaced are ones most leadership teams haven't yet been asked, which is the point. If your scores reveal large gaps in dimensions you haven't thought about, that's the assessment doing its job.
When you finish, two things happen automatically: the PDF generates for download, and the ROI calculator pre-populates with defaults reflecting your maturity scores. Both are immediate and free.