okikusan-public / articles / Beyond "Deployed but Unused" — Designing Internal AI Adoption
JA EN
← articles
CONTENTS12 sections
  1. 01Symptoms & problem
  2. 02The Claude phenomenon
  3. 03Three mechanisms
  4. 04Mandate vs voluntary
  5. 05KGI-KSF-KPI
  6. 06Second Brain
  7. 07Selling KPI consulting
  8. 08OSS alternatives
  9. 09Action by role
  10. 10Roadmap
  11. 115 failure patterns
  12. Summary
INTERNAL AI ADOPTION / 2026

Beyond "Deployed but Unused" — Designing Internal AI Adoption

Copilot rolled out company-wide, but usage sits at 10%. Onboarding ended with one briefing, and the winning prompts are buried in someone's personal Notion. This isn't a skills problem — it's a missing organizational design: there's no natural adoption loop.

How to reproduce a natural adoption loop with "AI Champion × monthly LT × hackathon × KPI linkage," covered from both the deploying side (IT / DX / line leads) and the provider side (AI product vendors).

AI adoption AI Champion KPI KSF Hackathon DX 2026.05.17 · 10 min read
FIG.0 — NATURAL SPREAD LOOP
// WHY CLAUDE CODE / CODEX SPREAD ACCELERATOR OSS release Individuals use it Post on social Reputation rises Fans accumulate New users flow in
The natural adoption loop of the social-media era. OSS becomes an accelerator by acting as free distribution of operational know-how. Inside a company, the three pieces of this loop (can try it / a place to post / posters get credit) are missing.
▍ THE PROMISE

"Deployed but unused" isn't a skills problem — it's a missing organizational design of a natural adoption loop. AI Champion × monthly LT × hackathon × KPI linkage can reproduce that loop inside a closed environment.

▍ TL;DR
§ 01 CONTEXT

The symptoms and what's really going on

>1-1Failure patterns we keep seeing

>1-2What's really going on

This is not a skill gap on the part of individual employees — it's an organizational design problem:

→ The fix is not pushing a product, but designing a natural adoption loop.

§ 02 OBSERVATION

Why did Claude Code / Codex spread on their own?

Before fixing internal AI adoption, observe the natural adoption happening outside. The reason Codex / Claude Code / Cursor exploded is that the five-step loop in Fig.0 kept turning — and the preconditions that keep it turning boil down to three.

On top of that, Codex / Claude Code open-sourced the CLI / Agent layer. This isn't just transparency — it's free distribution of operational know-how: how to drive an AI agent, how to run tools safely, how to design prompts. Dify and n8n follow the same shape.

▍ Three pieces of the natural adoption loop

(1) Individuals can try it (low barrier to experiment)
(2) There's a place to post (social / OSS / community)
(3) Posters get credit (followers / job offers / side income)

→ These three are absent inside companies. That's why it doesn't spread.

▍ PART 1 — THE DEPLOYING SIDE

The approach for the side rolling AI out internally

For IT / DX / middle managers / AI Champion candidates

§ 03 PART 1 / MECHANISMS

Design with three mechanisms

To reproduce the natural adoption loop inside a closed environment, combine three mechanisms.

FIG.1 — THREE MECHANISMS
FOUNDATION AI Champion Volunteer + recognition Start with 2-3 people // Sumitomo: ¥1.2B/yr saved FAST IMPACT Monthly LT 5-10 min each, fastest payoff Problem → Prompt → Impact // SmartNews: 196 attendees USECASE GENERATOR Hackathon Twice a year, on company time Prize + dev budget // Generates early adopters // Three pillars to reproduce the natural adoption loop inside a closed environment
Champion = foundation, LT = fast impact, Hackathon = use-case generator. The three only stick when combined.

>3-1AI Champion program (the foundation)

Success case: Sumitomo Corporation's "Copilot Champion" program → ¥1.2B/year cost reduction, 10,000 person-hours/month saved
Minimum viable setup: start with 2-3 Champions → full rollout in 3 months

>3-2Monthly use-case lightning talks (LT)

The mechanism with the fastest payoff.

>3-3Hackathon (the use-case generator)

Run study sessions = knowledge sharing, and hackathons = use-case generation, as a two-stage system.

TopicRecommendation
FormatHalf-day to one-day prototype on "solve a real work problem with AI" → demo
CadenceTwice a year (monthly is too much load, yearly is too thin)
WhenPrefer on company time. Saturdays skew participation toward singles and juniors
IncentivesWinner gets a dev budget / resources, top entries are considered for production
§ 04 PART 1 / BALANCE

★ Balancing mandate and self-direction

This is the biggest open question for internal AI adoption. Leadership wants to tighten with KPIs; the field wants to move on self-direction. How do you resolve that tension?

▍ The two failure modes at the extremes

Pure mandate: "everyone must post a monthly use case" → flood of low-quality posts, atmosphere sours
Pure voluntary: "whoever's interested can join" → only existing fans move, nothing spreads across the org

>4-1The Sumitomo Corporation model (recommended)

>4-2Design notes

§ 05 PART 1 / KPI

Measurement: KGI - KSF - KPI hierarchy

FIG.2 — KGI / KSF / KPI HIERARCHY
KGI / TOP-LEVEL GOAL Workflow efficiency & productivity KSF / Key success factors Champion Monthly LT Hackathon Wiki KPI / METRICS AI usage rate · Use cases shared · Time saved · Use cases in production
KSFs are success factors, not means to an end. KPIs are the tool for quantitative management of adoption. Run it as "a push," not enforcement.

>5-1Concrete KPIs and rough ranges

MetricRangeExample math
Minutes saved on meeting notes△2-5h / mo / person100 people × △3h × 12 months = △3,600 hours/yr
Meeting time reduction△10-20%tens of millions of yen/yr in overtime equivalent
Initial response time on inquiries△30-40%Customer support
Manufacturing / test cycle time△ x % by industryDesign the measurement per industry
Use cases postedx / monthSlack posts + LT talks combined
Use cases in productionx per quarterThings in production use (excludes PoCs)
▍ The quantification trap

Trying to put a number on everything brings on spreadsheet disease. Deliberately also pick up "things you can't measure but are valuable" (psychological distance between departments shrinks, juniors get more eager to post, etc.).

§ 06 PART 1 / AGGREGATION

Aggregate use cases into a "Second Brain"

The raw notes coming out of LTs, hackathons, and Slack should be collected raw at first. Don't aim for clean documentation up front.

FIG.5 — WRITE → STORE → READ
// 1. WRITE // 2. STORE // 3. READ pile up raw parallel preservation make sense in context Slack post LT slide Failure log Prompt Hackathon Notion memo WIKI // parallel Case A · context:sales Case B · context:eng Case C · same theme Case D · alternative + date + model [SALES] Pick Case A [ENG] Combine B + D [NEW HIRE] Learn from diffs // meaning-making happens at read-time, not write-time
Write = pile up raw / Store = parallel preservation + tags / Read = each reader interprets in context. The wiki doesn't classify — it stays findable.
▍ Artifact types that grow in the Wiki

cookbook (recipes by workflow) / prompt library / workflow templates / failure log / adoption checklist — raw material extracted and structured from Slack / Notion / Obsidian via Codex / Claude Code.

You could call this the organizational version of Karpathy's LLM Wiki.
→ See Obsidian → LLM Wiki → HTML → AI Deploy for a separate write-up.

>6-1Handling duplicates and disagreements

Keep aggregating long enough and it will happen: multiple teams post different prompts for the same workflow, one insists "this is the right one" while the other has a better alternative, the same prompt yields different results in different teams. The rule is "keep both with context," not "pick a winner and delete the other."

SituationCommon failureRecommended approach
Duplicates (same case from multiple teams)Delete the older one and overwriteKeep both and cross-link. Champion decides on consolidation each quarter
Different approachesSnap-judge a winner and delete the otherShow as "Case A / Case B" with context attached (department, data scale, constraints)
Same prompt, different resultReport "doesn't work" and stop thereRecord under "Known variations." Often the most valuable information
▍ Why "parallel preservation" is the right default

Organizational use cases are context-dependent. Sales and engineering can legitimately have different "right answers" with the same tool. Calling one "best practice" excludes the other and kills the urge to post. Show all the evidence; let the user decide — this is the same principle whether the Wiki is personal or organizational.

One step deeper: even the same person uses a tool differently depending on context. So the wiki's job is less "build the right classification" and more "pile up the raw material so each reader can make sense of it in their own situation." Meaning-making happens at read-time, not write-time — a Champion's curation is less about organizing and more about keeping things findable.

>6-2Operational mechanisms

Lightweight on classification, heavy on findability.

▍ PART 2 — THE PROVIDER SIDE

The approach for the side providing AI products

For SaaS vendors / AI product providers / CSMs / solutions engineers

§ 07 PART 2 / PROVIDER

Product alone won't sell: shift to selling KPI consulting

If the provider just "builds a good product and sells it," the client falls into three pains:

at the same time.

>7-1Co-design the client's KGI / OKR

Break the provider-side KGI (GMV growth) into a product of three factors.

FIG.6 — KGI / GMV DECOMPOSITION
// PROVIDER-SIDE KGI DECOMPOSITION ▍ KGI GMV growth ↑ = decompose × × Customer accounts // sales / marketing market reach Users per account // adoption rate penetration Tokens per user // usage depth depth // product of last 2 = use-case density ▍ PROVIDER'S LEVERAGE Use-case density ↑ // Bake Part 1's Champion / LT / Hackathon into CS to move the needle
Provider-side KGI factors into three multiplicands. The product of the last two = use-case density — the leverage point the provider can actually move. Don't push license counts; lift depth of use per account.

So the provider's plays all converge on "maximize use-case density." Not pushing license counts — the growth headroom in GMV is depth of use per account. That's why the provider should bake Champion / LT / hackathon (from Part 1) into the service itself:

These can be built in as part of the contract.

§ 08 PART 2 / ALTERNATIVES

OSS / community strategy: borrowing trust forward

SaaS / API vendors release OSS not out of altruism but to borrow trust forward.

TacticWhat it gets you
OSS releaseShow "easy to switch off" and kill the fear of lock-in
Public templatescookbook / prompt library / workflow templates (distributing know-how)
API / SDK accessLets people learn integration patterns
Community opsUser groups / Discord (build a posting culture through peer-to-peer)
★ Host hackathonsA loop where the market discovers use cases for you

In particular, hackathons are powerful as places where "users discover use cases themselves." You can run internal community + external hackathons in parallel.

▍ COMMON

Things to move on tomorrow

Execution guide common to deploying side and provider side

§ 09 COMMON / ACTIONS

First action by role

What matters is not what you read but what you do. Bring it down to the level of "doable today or this week":

FIG.3 — FIRST ACTION MATRIX
[EXEC] EXECUTIVES At the next All Hands, say "AI experiments OK, failure OK." Reserve budget for a Champion program. [MGR] MIDDLE MANAGERS Ask 1-2 AI-savvy people on your team: "Want to be a Champion candidate?" [STAFF] INDIVIDUAL CONTRIBUTORS Post one recent use of AI to the #ai-use-cases channel on Slack/Teams (a paragraph is enough) [CSM] PROVIDER CSM Book a call with your client to interview their current KPI design // THINGS YOU CAN DO THIS WEEK
Don't stop at reading — translate to a concrete action by role this week. Four axes: executives / middle managers / individual contributors / providers.
§ 10 COMMON / ROADMAP

3 months + alpha roadmap

FIG.4 — 12-MONTH ROADMAP
M1 FOUNDATION
Pick 2-3 Champions
Open Slack channel
All Hands declaration
Initial KPI definition
M2 BROADCAST
Start monthly LT
Build the template
Spin up the Wiki
First KPI snapshot
M3 GENERATE
Run hackathon
Winners in newsletter
1 case in mgr review
Build cross-team ties
M4-6 EXPAND
More Champions
Build the cookbook
Cross-team LT
Tune the KSFs
M7-12 STICK
Annual hackathon
External talks
Cross-team use cases
Market formation
// 3 MONTHS TO LIFT-OFF → 12 MONTHS TO STICK
Five phases: foundation → broadcast → generate → expand → stick. The key to sustainability is "never miss a monthly LT."
§ 11 COMMON / FAILURES

Five failure patterns

First, here are the three tension axes we've been discussing, on one page. Most failure patterns come from going all the way to one side of one of these axes.

AxisLeft: fails if pushed all the wayRight: fails if pushed all the wayThe right answer
Mandate vs self-drivenSqueeze with KPIs → self-direction diesAll-voluntary → nothing happensLight KPI linkage + evaluation incentives
KPI vs aggregationChase numbers only → spreadsheet diseaseJust collect → impact never reaches leadershipShow via KPI, accumulate as knowledge
SaaS vs OSSSaaS only → lock-in / cost bloatOSS only → ops overhead / knowledge silosCompete on use-case density (tools are means)

The typical pitfalls to avoid:

  1. Running KSFs as a mandate: hard KPI enforcement → self-direction dies
  2. No posting culture: presenting doesn't get rewarded → no one presents — a vicious cycle
  3. Leadership doesn't say "experiments are OK": zero psychological safety, no one tries anything
  4. Use cases never get aggregated: events get hyped, but good prompts scatter across Slack and personal Notion and never become org knowledge
  5. No KPI story on the provider side: adoption cost and impact aren't linked, procurement stalls
▍ THE WORLDVIEW — FROM "FORCED" AI TO AI PEOPLE ACTUALLY USE

Design the natural adoption loop from both sides: the organization and the provider

The core of internal AI adoption is not pushing a product, but designing a natural adoption loop. The four pieces to line up:

  1. A place to post (Slack channel / LT / Wiki)
  2. A culture where posters get credit (Champion program / executive presentations)
  3. A way to try things small (PoCs / hackathons / start with non-confidential data)
  4. KPIs that make the impact visible (deliver time and cost savings to leadership)

On top of that:

  • Balance mandate and self-direction (light KPI linkage + positive incentives)
  • Run KPI and aggregation in parallel (show via numbers, accumulate as knowledge)
  • Compete on use-case density, not SaaS-vs-OSS (tools are just means)

Don't order people to "use AI." Instead, both the organization and the provider should set up a design where using AI gets you credit, is fun, and makes tomorrow's work easier. That's the core of internal AI adoption from 2026 onward.