If you're new, read in this order
- Read Part 0 and take the paper's four findings in five minutes — Why buying the accounts isn't enterprise AI adoption
- In Part 1, count your "don't know"s on the 12-question self-check — four or more means you're still at individual use — The 12 questions a company must answer after adopting AI
- Go to the layer where your "don't know"s cluster. Permissions → Layer 2; where to put it → Layer 3; worried about incidents → Part 9; want the effect in numbers → Part 8.
- Pick one first seat (Parts 4 and 11) and turn it into a document together in a ₩250K scoping pass. The contract comes after you've seen the demo.
Layer 1 · Visibility — how AI is actually being used in our company
Without measurement you stop at "it seems to be going well." The paper's four metrics, MAEUM's fifth (retention), and the one page the CEO looks at every week.
+2 more in this layer — Korean edition →
Layer 2 · Governance — who is allowed to do what
Permissions, data, security, risk. This is the layer where incidents happen, and the one most companies skip.
+3 more in this layer — Korean edition →
Layer 3 · Workflow — where AI enters the work
Put it everywhere at once and it fails. One first seat, first seats by department, rollout order and resistance, meetings, email and chat.
+4 more in this layer — Korean edition →
Layer 4 · Execution — what AI actually does, and who checks it
How far AI, from where humans. One agent, then several. And how to evaluate results in numbers.
+3 more in this layer — Korean edition →
Layer 5 · Knowledge and models — what AI knows, and which model answers
Connect company memory (RAG), organize documents so they can be found, turn prompts into company assets, keep models swappable.
+4 more in this layer — Korean edition →
Layer 6 · Cost and results — what it costs, what's left
Four places money leaks, ROI calculation, per-seat subscriptions vs flat fees. We don't hide our prices.
+1 more in this layer — Korean edition →
Layer 7 · Org design — how people and agents are divided
When the division of labor changes, the org chart changes. New hires are the deepest users, and decisions must be recorded to become the basis for the next one.
+3 more in this layer — Korean edition →
Layer 8 · Redesign loop — do you keep re-shaping the organization from results?
What the paper called "slow co-invention" is repetition. A one-hour monthly review is what the loop actually looks like.
+2 more in this layer — Korean edition →
Why we wrote this series
AI companies sell models. MAEUM builds the place the model sits. As the paper says, models are becoming commonplace and value is moving to the integration layer — there's no reason to hide how that layer is designed. You're welcome to use this series as-is (with attribution). If you try it yourself and need hands, that's what the ₩250K scoping pass is for. At the end of every part we separate "what the paper says" from "MAEUM's extension" — read it knowing where the evidence ends and our design begins.
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 I have to read all 26 parts?
No. Read Parts 0 and 1, then go to the layer where your "don't know"s cluster on the 12-question self-check. The rest is a reference you open when you need that layer.
What is this series based on?
A working paper by OpenAI researchers with professors from Columbia and Wharton, "How Organizations Use AI: Evidence from ChatGPT" (2026-08-11) — an analysis of actual usage records from 1,764 ChatGPT Enterprise customers and 17.4 million messages. The last section of every part separates what the paper says from MAEUM's extension. MAEUM is neither an author of the paper nor an OpenAI partner.
Does this only apply if we buy ChatGPT Enterprise?
No. The paper's data happens to be ChatGPT Enterprise, but the design questions (permissions, division of labor, workflow, memory, evaluation) are the same whichever AI you use. MAEUM doesn't sell a specific AI; we build structures where the business keeps running when the model changes. If AI isn't needed, we don't add it.
We're a 10-person company — does a large-enterprise paper apply?
Don't transfer the numbers directly; the direction is the same. In the paper, large companies adopt first but per-employee usage after adoption is actually lower — spreading takes longer. In a small company that time is short. From Layer 3 onward it isn't a game of company size.
Can we use the series as internal material?
Yes. With attribution (the MAEUM blog and the OpenAI paper) you may use it freely. Append index.md to any article URL for the markdown source.
What does it cost?
Scoping pass ₩250K (5 sessions, up to 90 min each, working demo included). SaaS (managed): build from ₩700K + ₩490K / ₩990K / ₩1.49M/mo (no per-seat pricing; improvements, new features and a monthly report included). SI (ownership): build from ₩2.5M (full build from ₩7M+), care pass ₩490K. VAT excluded; final price confirmed after scoping. The working demo before any contract is free.
