Five questions. Five documented ways enterprise AI breaks.

Before the check

First: five questions no organization can answer today.

The check below scores the documented ways enterprise AI breaks. These five make it personal — the answers don't exist anywhere in your organization yet, because nobody has decomposed work below the job description.

1. “Your AI plan will automate thirty percent of the work. Thirty percent of what?”
2. “Pick any task. Show me the evidence it belongs to a person, a machine, or both.”
3. “Which decisions is AI making in your organization right now — and who signed for each one?”
4. “What does one unit of that work cost by hand, versus by machine — and how do you know?”
5. “What work must stay human — and where is that list?”

Why each one is unanswerable today →

Leave this empty:

1 · The 95% problem. Where do your AI initiatives actually stand?

No real initiatives yet Pilots running, nothing measurable Tools deployed, ROI unclear Measurable P&L impact

Why we ask: MIT's 2025 GenAI Divide study found 95% of enterprise GenAI pilots deliver zero P&L return, despite $30–40B invested.

2 · The redesign gap. When AI arrived, did the way work is divided actually change?

AI layered onto existing workflows Some roles adjusted informally We redesigned some workflows We formally re-divided work between people and AI

Why we ask: McKinsey found workflow redesign is the single biggest driver of bottom-line impact from AI — yet only about 21% of organizations have done it.

3 · The accountability problem. If an AI made a consequential mistake in a governed process tomorrow, could you show exactly what it did and who approved it?

Yes — full audit trail Partially — logs exist somewhere No AI isn't allowed near governed work

Why we ask: Gartner predicts over 40% of agentic-AI projects will be canceled by end-2027 — chiefly for unclear value and inadequate risk controls.

4 · The shadow-AI problem. How confident are you that you know where AI already touches your sensitive work?

Confident — inventoried and governed We suspect ungoverned use We know it's happening, unmanaged We haven't looked

Why we ask: 2025 studies put unapproved AI use between half and four-fifths of employees — and IBM found breaches involving shadow AI cost roughly $670K more.

5 · The allocation blindness problem. Could you say today, in numbers, which parts of one governed process should be automated and which must stay human?

Yes, we've quantified it Opinions, not numbers No — that's exactly the question We haven't framed it that way

Why we ask: this is the question NOW OS computes — every atomic action scored against a published rule into one of four honest modes.

Work email (so we can connect this to your demo request) Organization

See where you stand

Submitting opens a pre-filled email to info@nexusofwork.com with your answers — nothing is sent until you hit send in your mail app.