The Complete Guide to Drafting Individual Support Plans with AI | How to Cut Service Manager Time in Half

Hello, this is Miya (@employment38).

I work in marketing at a Type B Continuous Employment Support facility in Sapporo.
I work in marketing.

Today, I am going to dive into the
most important document in the welfare field.

The Individual Support Plan.

If you are a Service Management Responsibility Manager (Sabikan),
doesn't just hearing this term
make your shoulders feel heavy?

In this article,
I will explain the specific steps
to create a draft of an Individual Support Plan
using AI.

Are you just 'making' your Individual Support Plans?

Type B Continuous Employment Support facilities
have an obligation to create
an Individual Support Plan for each and every user.

Long-term goals, short-term goals,
the individual's efforts, support content, and achievement status—.

It is truly the blueprint for support
to help sustain the 'life that is unique to that person' for the user.

But, to be honest about the reality on the ground.

I believe there are quite a few Service Managers and administrators
who feel this way.

If you have 20 users,
you create plans for 20 people,
monitor them periodically,
and review them as necessary.

And all of this while balancing
daily support tasks and staff management.

Because there is no time,
you reuse templates with minor wording changes.
You just make slight adjustments to a copy-paste from last time.

Before you know it,
the plans for every user
end up looking the same—.

There might be some of you
who have experienced this.

The Individual Support Plan is
a document that is also heavily checked during on-site inspections.
If there are deficiencies, there is a risk of payment reductions.

That is precisely why
the focus tends to be on getting the format right,
and the original purpose of 'seriously thinking about a plan unique to that person'
ends up being put on the back burner.

That is the theme of this article.

Prerequisite: AI creates a 'draft'

Let me make this clear from the start.

AI cannot 'complete' an individual support plan.

What AI can create is
strictly a 'draft' (a rough sketch or starting point).

An individual support plan is created by
synthesizing the user's own intentions,
the results of the assessment,
and the professional judgment of the support team.

The person who knows the user's 'uniqueness'
is you, the support worker who interacts with them every day.
It is not AI.

Therefore, the role of AI is this:

The time spent staring at a blank Word file,
wondering 'where should I start?'
The time spent formatting the text.
The time spent searching for appropriate expressions.

Leave this part to AI.

And then, the support worker
reads the draft produced by the AI
and thinks as follows:

• 'Does this goal align with the individual's intentions?'

• 'Is this support content sufficient?'

• 'Is there a more fitting way to express their personality?'

In other words, AI is not a tool to take away 'thinking time',

but a tool to create 'time to focus on thinking'.

Overall Flow: 5 Steps

Drafting an individual support plan with AI
proceeds in the following 5 steps.

Step 1: Organize information (10 minutes)
Organize interview notes, assessment information,
and reflections on the previous plan
into bullet points.

Step 2: Have AI create a draft (2 minutes)
Input the organized information along with the prompt into
NotebookLM.

Step 3: Review and revise the draft (15 minutes)
Read the AI output and
verify facts and revise the content.

Step 4: Review with the support team (as usual)

Step 5: Explain to the individual and their family and obtain consent (as usual)

Steps 4 and 5 are the conventional processes.
AI shortens the parts in steps 1 to 3.

Step 1: Organizing information (this is the most important part)

The trick to getting AI to create a good draft is,
as I mentioned in the first installment,
**"providing specific information."**

For an Individual Support Plan,
organize the following information in bullet points before
giving it to the AI.

【利用者の基本情報】※個人を特定できない範囲で
・年代：
・性別：
・主な障害種別（大まかに）：
・利用期間：

【本人の意向・希望】
・本人が言っていること、やりたいこと：
・将来の目標や夢（あれば）：

【現在の状況】
・得意なこと、できていること：
・課題になっていること：
・体調面で配慮が必要なこと：
・対人関係の様子：

【前回の計画の振り返り】
・前回の目標の達成状況：
・うまくいったこと：
・うまくいかなかったこと：

【支援チームの意見】
・担当支援員が感じていること：
・他のスタッフからの情報：

It does not need to be perfect prose.
A list of keywords is also fine.

Since AI only creates text based on the
"information provided,"
the quality of the input
directly translates to the quality of the output.

Step 2: Have AI create a draft

Input the organized information
along with the following prompt
into Gemini.

あなたは就労継続支援B型事業所の
経験豊富なサービス管理責任者です。
以下の情報をもとに
個別支援計画書のドラフトを作成してください。

【作成ルール】
・長期目標を1つ設定する
 （6ヶ月〜1年の期間で達成を目指す内容）
・短期目標を2〜3つ設定する
 （3ヶ月程度で達成を目指す具体的な内容）
・各短期目標に対して
 「本人の取り組み」と「支援内容」を記載する
・目標は本人の意向を反映した前向きな表現にする
・支援内容は具体的で
 誰が読んでも同じ対応ができるレベルにする
・専門用語は最小限にし
 本人やご家族が読んでも理解できる表現にする
・文体は「ですます調」で
 あたたかみのある表現にする

【利用者の情報】
ここにステップ1で整理した情報を貼り付ける

Input example

【利用者の情報】
・年代：30代
・性別：男性
・主な障害種別：精神障害（気分障害）
・利用期間：約1年

【本人の意向・希望】
・「もう少し長い時間、集中して作業したい」
・将来的にはA型事業所への移行も視野に

【現在の状況】
・データ入力作業が得意で、正確性が高い
・午前中は集中力が続くが
 午後になると疲れが出やすい
・他の利用者とは穏やかに関われている
・体調に波があり、月に2〜3回欠席がある

【前回の計画の振り返り】
・前回の目標「週3回以上の安定した通所」は
 ほぼ達成（月平均3.5回/週）
・午後の作業で集中が切れたとき
 自分から休憩を申し出る場面が増えた
・朝の体調確認シートの記入が定着した

【支援チームの意見】
・担当支援員：体調管理の自己認識が向上している。
 作業の幅を広げるタイミングかもしれない
・生活支援員：服薬管理は安定。
 睡眠リズムの乱れが欠席につながるパターンがある

AI output image

Based on this information, NotebookLM
will output a draft with the following structure.

<Long-term goal>
While stabilizing physical condition,
gradually increase work hours,
and proceed with preparations
with an eye toward transitioning to a Type A facility.

<Short-term goal 1>
Increase the number of days able to work through both morning and afternoon

• Individual's efforts: Perform 5 minutes of stretching before starting afternoon work to consciously switch gears.

• Support details: Focus afternoon work on data entry, which the individual excels at, and provide verbal prompts if their concentration wanes.

<Short-term Goal 2>
Improve self-management skills for physical condition

• Individual's efforts: In addition to the daily physical condition check sheet, start recording sleep hours.

• Support details: Review physical condition and sleep during a weekly mini-interview, and brainstorm preventive measures for absence patterns together.

<Short-term Goal 3>
Challenge new tasks and expand the range of what can be done

• Individual's efforts: Try a task never experienced before once a month.

• Support details: Propose new tasks that match the individual's interests and characteristics, and reflect their feedback after the experience in the next plan.

Of course,
this is not a finished product ready for immediate use.

This is where the support worker's job begins.

Step 3: Review and Edit the Draft

Review and edit the draft generated by AI
from the following perspectives.

① Is the individual's intent accurately reflected?
AI is good at processing the information provided
to make it sound plausible.
However, check if it deviates from
what the individual truly desires.
Please confirm while recalling the words spoken during the interview.

② Are the goals specific and measurable?
Instead of "aim for stable attendance,"
use expressions that allow you to judge whether it was achieved,
such as "aim for attendance 4 or more times a week."
If the AI output is abstract,
please break it down into numbers or behavioral levels.

③ Are the support details specific and reproducible?
Instead of "provide appropriate prompts,"
use specific details such as "check facial expressions and posture at 2:00 PM,
and if fatigue is observed,
suggest a 10-minute break," to ensure
the response is consistent regardless of who provides it.Is it specific enough?

④ Are the individual's strengths reflected?
Is the focus only on challenges?
Check if the goal setting is based on
"what they can do" and "what they are good at."
Since AI tends to create problem-solving-oriented text,
please check this consciously.

⑤ Is there continuity with the previous plan?
Has the level increased too suddenly?
Or conversely, has it regressed?
Check if the settings are incremental,
based on the achievement status of the previous plan.

⑥ Is the expression understandable to the individual and their family?
Are there any technical terms?
Is the tone condescending?
Instead of "instruct," use "think together."
Instead of "make them improve," use "support their efforts."
Such consideration for expression
is an area where AI still struggles.

⑦ Are there any personal information issues?
Reconfirm the rules explained in Part 1.
Did you use real names when inputting into the AI?
Does the output result contain
information that could identify an individual?

Are there any legal issues with using AI to create plans?

I think many people are concerned about this point.

Let me give you the conclusion.

The responsibility for creating the Individual Support Plan
ultimately lies with the Service Management Responsibility Manager.

There are no legal restrictions
on what tools you use
in the creation process.

However, please observe the following points.

• Ensure that user information entered into AI is anonymized so that individuals cannot be identified.

• Do not use AI output as-is; always review and revise the content as a professional.

• Explain the content of the plan to the individual and their family and obtain consent in person as usual.

As long as these are followed,
I believe that utilizing AI will actually
work to improve the quality of the plans.as well.

This is because
the time saved on "writing text"
can be dedicated to "thinking about the content."

Application: Utilizing AI for Monitoring as well

Once you are able to create Individual Support Plans,
next,try utilizing AI for monitoringas well.

The method is basically the same.

あなたは就労継続支援B型事業所の
経験豊富なサービス管理責任者です。
以下の個別支援計画書と
直近3ヶ月の支援記録の要点をもとに
モニタリング報告書のドラフトを作成してください。

【作成ルール】
・各短期目標の達成状況を
 「達成・一部達成・未達成」で評価する
・達成状況の根拠となる具体的な事実を記載する
・次期計画に向けた課題と提案を記載する
・本人の意向の変化があれば反映する

【現在の個別支援計画書の内容】
ここに現在の計画書の内容を貼り付ける

【直近3ヶ月の支援記録の要点】
ここに箇条書きで要点を貼り付ける

Since monitoring is a task of
"comparing the previous plan with the current situation,"
it is very compatible with AI.

Just by having it draft
the evaluation text for achievement status and
the proposal text for the next plan,
you can save a significant amount of time.

Summary: AI is a partner that creates "thinking time" for Service Managers

Finally, I will repeat
the important point one more time.

AI is not a tool to
"complete" an Individual Support Plan.

What AI excels at is
organizing information and building the framework of the text.

What humans (support staff) excel at is
breathing "individuality" into that framework.

Please try this out
the next time you create a support plan.

Notice

If you haven't read Part 1 yet,
↓
[Copy-Paste OK] How to Save Time on Welfare Documentation with Gemini | AI Introduction for Welfare Professionals

Part 2 is here
↓
Want to Introduce AI but Staff Aren't Interested? | How to Overcome the "It's Too Early" Barrier at Type B Facilities

Part 3 is here
↓
Not Getting Enough Tour Requests? | Website SEO for Type B Facilities | Introduction to Web Attraction for Welfare Facilities

If you found this article helpful,
it would be encouraging if you could press the "Like" button. If you
follow me, you will receive notifications for the latest articles.

[About Me]

Practicing and sharing AI x Web utilization for welfare and employment support facilities.
Registered Nurse => Web Industry => Struggling as a marketing lead for employment support.
I am also publishing content on how to halve work time using Claude, Gemini, NotebookLM, and Typeless, as well as methods to increase tour requests through Web x SEO.

My Profile

The future I want to create with Welfare x AI/Web

[Social Media Accounts]

[Post Category List]

AI Utilization Techniques for the Welfare Field

Complete Guide to Web Attraction for Welfare Facilities

AI Utilization Techniques for the Welfare Field