MAEUM BLOG · AI adoption

The 12 Questions a Company Must Answer After Adopting AI — The Problem Isn't ChatGPT

2026-09-08 · MAEUM — an AI engineering company that publishes its prices
The real problem in enterprise AI isn't "can we access AI" but "is AI integrated into the company". The moment you buy ChatGPT, twelve questions remain exactly where they were — who uses it · for what · which data goes in · which model · how far AI acts on its own · who reviews · can it operate real systems · who approves the result · who is accountable when something goes wrong · how do we know the result was good · what does it cost · how do we improve next time. Four or more "don't know"s and AI stays in personal accounts. MAEUM fills in all twelve with you inside a ₩250K scoping pass, then plants the answers in a system — SaaS (managed) build from ₩700K + ₩490K / ₩990K / ₩1.49M/mo.

Why twelve — what the paper says, and what we added

OpenAI's 2026 paper "How Organizations Use AI" sums up enterprise AI adoption in one sentence: "Adoption is only the beginning of deployment." After buying AI, a company has to find where to use it → invest in complementary capabilities → redesign workflows → integrate it stably into daily work. (The paper is summarized in Part 0 of the series.)

That's what the paper says. To apply those four stages to a real company, questions arise that have to be answered. We distilled the ones we run into on the ground every time into twelve. These twelve are not the paper's claim but MAEUM's extension — the paper said "integration is needed," and we write "to integrate, you have to decide these."

The 12 questions — which design area each belongs to

#QuestionDesign areaWhat happens without an answer
1Who will use itPermissionsAnyone uses it, or no one does
2For whatTask mappingStuck at "what should we do with AI"
3Which data may go inData governanceCustomer personal data flows into personal accounts
4Which modelModel routingExpensive model for simple classification, cheap model for contract review
5How far does AI act on its ownAutonomy scopeThe owner finds out later about a text the AI sent
6Who reviewsHuman reviewWrong answers go straight to customers
7Can AI operate real systemsExecution rightsOrders and refunds go out with no human check
8Who approves the resultDecision authorityDecisions with no owner pile up
9Who is accountable when it failsAccountability structureAfter an incident it ends with "the AI did it"
10How do we know it was goodEvaluationEffects described only as feelings
11What does it costCost managementSurprise at the month-end bill
12How do we improve next timeOrganizational learningThe same mistake repeated department by department

Each of the twelve is one part of the series. Permissions (1·7·8) → Part 2; autonomy and review (5·6) → Part 3; task mapping (2) → Part 4 (Korean); model (4) → Part 5 (Korean); data (3) → Part 6 and Part 9 (Korean); evaluation (10) → Part 8 (Korean); cost (11) → Part 10 (Korean); improvement (12) → Part 12 (Korean).

Self-check — count your yes / no / don't know

Next to each of the twelve questions, write one of yes · no · don't know. "No" is fine — it means you know nothing has been decided. The problem is "don't know": nothing was ever decided, and you don't know who should decide.

"Don't know" countStageWhere you are
0–3Integration in progressAI is inside the workflow; only the gaps need filling.
4–7Individual-use stageEmployees use it individually, but there's no company-level design. The paper's "adopted but not deployed" state.
8–12Pre-adoptionEven with accounts bought, you're effectively pre-adoption. It's faster to decide everything together from the start.

Most small and mid-sized companies are in the middle row. Nothing to be embarrassed about — the authors write that the paper's 1,764 large companies are also "still learning where it belongs."

An accounting-firm scenario — how the 12 questions actually come up

Say a six-person tax and accounting firm decides "let AI handle the reminders for clients who haven't submitted their bookkeeping documents." The twelve questions land like this.

Writing those twelve lines takes 90 minutes. That's what the first scoping session actually looks like.

The order for filling in all 12 within five scoping sessions

MAEUM's ₩250K scoping pass (5 sessions, up to 90 min each) is structured to fill in these twelve questions.

SessionQuestions filledOutput
11 Who · 2 For whatList of repetitive tasks + one first seat chosen
23 Data · 4 ModelList of data allowed in, model selection criteria
35 Autonomy · 6 Review · 7 Execution · 8 ApprovalPermission table (the 20-permission matrix from Part 2)
49 Accountability · 10 Evaluation · 11 CostNamed owners, 3 metrics, monthly cost cap
512 Improvement + demoA working demo (free) — the contract comes after

When the five sessions end, the twelve answers remain as a one-page document. Even if you don't sign, that document is yours.

Sources — what the paper says / MAEUM's extension

MAEUM list prices — VAT excluded · starting prices · final price confirmed after scoping
Scoping pass (5 sessions)₩250K (≈ $179) · up to 90 min each · working demo included
SaaS · managed (recommended)Build from ₩700K (≈ $500) + ₩490K / ₩990K / ₩1.49M/mo (≈ $350 / $707 / $1,064) — servers, AI, improvements, new features and a monthly report included. No per-seat pricing.
SI · ownershipBuild from ₩2.5M (≈ $1,786); full build from ₩7M+ (≈ $5,000+) — no monthly fee afterwards, care passes when needed

KRW is authoritative; USD figures are approximate. SEO and AI-search visibility work is included in every build at no charge. The demo is free — try it, and sign only if you like it.

Frequently asked questions

Do all 12 have to be decided before we start?
No. Decide 1 and 2 (who, for what) and one first seat can start. The rest get filled in while that first seat runs. But 3 (data) and 7 (execution rights) must be decided before starting — that's where incidents happen.
We have no IT person. Who decides these?
It isn't IT's job. 1·2·8·9 are the owner's; 3·5·6 belong to whoever does that task. For 4·7·10·11 we give you a draft and the owner confirms. The ₩250K scoping pass is where that happens.
Doesn't the ChatGPT Enterprise admin console handle all this?
It handles 1 (who) and part of 11 (cost). The other ten, no AI vendor's console decides for you. Only someone who knows the company's workflow can.
If we settle the 12 answers, can we switch AI vendors?
That's the core benefit. The twelve answers aren't tied to a specific AI. When the model changes, only 4 (which model) needs updating; the rest carry over. Part 5 covers this in detail.
What does it cost?
Scoping pass ₩250K (5 sessions, up to 90 min each, working demo included). Planting it in a system afterwards: SaaS (managed) build from ₩700K + ₩490K / ₩990K / ₩1.49M/mo (no per-seat pricing); SI (ownership) from ₩2.5M. VAT excluded.

See it working before you decide

Tell us what you do and what you need. We'll show you a working demo first — the demo is free.