🐙Claude Learning Log: Part 4 - Full Disclosure of 'Automated Spot Consulting Opportunity Search & Application Drafting' Skills! It's Super Convenient
The story of how I got Claude to give me back my 'Monday mornings'
Hello, this is Roel.
Every Monday morning.
I used to make it a habit to browse through multiple spot consulting platforms.
▼ What is spot consulting?
Opening new project listings one by one, determining if they fit my background, and writing application drafts. If I'm a match, it's 20,000 to 30,000 yen for an hour.
As a way to 'earn some pocket money in my spare time' since becoming independent, it had become a non-negligible source of income. However, to be honest, this task was a hassle. Opening the project, reading it, cross-referencing it with my background, thinking about the application, writing it, and sending it. It took 10 minutes per project. If I did that across 4 or 5 sites, my morning would basically disappear. I felt somewhat unsatisfied with the structure of sacrificing my sharpest time on Monday mornings just for 'pocket money'.
🛠 Giving AI my 'outline'
So, I created a Claude Code skill for this.
Every Monday at 9 AM, it launches automatically. An AI agent browses the spot consulting sites I usually check, extracts new projects, and even writes the application drafts. All I do is review what comes out and give the OK.
One hour has turned into 10 to 20 minutes.
▼ Recommended spot consulting services
If it stopped here, it would just be another common story about 'automating with AI.'
But the real core of this skill lies elsewhere.
Within the skill, I have fed it my CLAUDE.md, which is my 'user manual'.
20 years of organizational experience, my career trajectory as Sales, Marketing, CMO, New Business, and AI Department Head, my first year since going independent, my areas of expertise, areas I'm not good at, market rates for compensation, and my habits in writing applications.
I explained in great detail how I write my 'user manual' in the previous part of this series.
💡 Claude Learning Log Part 3: Introduction to 'CLAUDE.md' 🤖 ~Full disclosure of the note rules section~
In other words, when the AI filters projects, it does so while knowing 'the person called Roel.' Therefore, the candidates that come out are not just projects that seem to fit, but projects judged with the resolution of my background. The application drafts are not just filled-in templates, but come out in a writing style that sounds like how I would write if I did it carefully.
In terms of accuracy, honestly, it's no different from me doing it myself. Sometimes, the AI's first draft is even sharper than the application I would write with a groggy head on a Monday morning.
▼ Sharing an example with some original text redacted ▼
==========================================================
Spot Consulting Weekly Report — 2026-05-04 (Mon)
v2 filter applied version. Scoring in the order of: Exclusion Gate (Blacklist) -> Whitelist Bonus -> Threshold (Top Pick 85+ / Candidate 75-84 / Reference 60-74 / Exclude <60) -> Final 5-Question Gate.
This week's summary
Platforms browsed: 3/4 sites (●●● skipped due to Monday's regular Offline for Maintenance)
Total scanned projects: Approx. 60 projects (●●● 3 pages) + 83 projects (●●● 3 pages) + 2 projects (●●● spot) = Approx. 145 projects
Top picks (85+): 2 projects
Candidates (75-84): 3 items
Reference (60-74): 3 items
Many Blacklist Exclusions: Most were filtered out due to being PR/IR only, market structure overview types, active practitioner requirements, or industry restrictions (medical/construction/manufacturing/finance/agriculture/semiconductor/infrastructure).
🥇 This Week's Top Pick (85+)
🥇1st Place | [Seeking ●●● Marketers] Request for User Interview regarding Generative AI x Marketing Utilization and Content Strategy (91 points / ●●●)
Score Breakdown: Experience Match 24/25, Compensation 8/25, Specificity 22/25, Appeal 22/25 (Whitelist 'AI x Marketing' +10, AI context +5)
Compensation: 20,000 JPY
Time: 60 minutes
Format: Online
Company: ●●● Co., Ltd.
Industry: ●●● / Tokyo
Application Deadline: Until 2027/03/31
Reason for Recommendation: The project title perfectly matches 'Generative AI x Marketing.' It is right in the middle of my current main focus (●●●). Although the compensation is low at 10k, the client has posted multiple projects on the same theme, indicating continuity. It is highly valuable to take for reference stocking purposes.
▼ Application Text (for copy-paste)
I can speak to the reality of working in the field from the position of in-house support for Generative AI x ●●● marketing.
🥇2nd Place | Want to interview about the launch and operation of a corporate AI training business (96 points / ●●●)
Score Breakdown: Experience Match 22/25, Compensation 13/25, Specificity 22/25, Appeal 24/25 (Whitelist 'AI Training Launch' +10, AI Context +5)
Compensation: 15,000 JPY - 20,000 JPY (excluding tax)
Time: 1 hour
Number of Openings: 1 person
Format: Phone/Web Conference
Company: Company name disclosed after matching (●●● industry)
Application Period: 2026/05/01 - 2026/05/08
Reason for Recommendation: The title 'Launching an AI Training Business' is exactly my current work. Since it is a single-person opening, the probability is high if there is a match. ●●● industry client = potential for relationship building through old connections.
▼ Application Text (for copy-paste)
As someone who has 'launched' a corporate AI training business, I can speak to everything from zero-to-one design to revisions after one year of operation.
🥈 This week's candidates (75-84)
🥈 3rd Place | [Interview] Over 5,000 employees | Inner Branding Design Advisor for translating mid-term management guidelines into 'organizational culture' (82 points / ●●●)
Score Breakdown: Experience Match 18/25, Compensation 22/25, Specificity 19/25, Appeal 17/25 (●●● name recognition +3, high unit price +3)
Compensation: 30,000 JPY - / negotiable
Format: Online
Company: ●●● Co., Ltd.
Application Deadline: Until 2027/04/30
Reason for Recommendation: Translating a large corporation's mid-term management policy into organizational culture. My strength lies in the adjacent area of CMO (Marketing x Organization x Branding). It's an 'advisor' role, not just PR/IR—meaning it's a sounding board type, and it does not fall under the v2 blacklist. High unit price starting at 30,000 yen.
▼ Application Text
I can speak from the position of an advisor who accompanies large-scale organizational culture transformation from a marketing perspective. When I was an executive officer at ●●● (previous job), I designed a consistent flow from vision penetration to action guidelines to KPIs, improving employee NPS by ●●●. At ●●● (job before last), I also implemented a three-layer framework for translating management policies into employee handouts, training, and evaluation systems. I can help from the perspective of designing inner branding not as something confused with corporate PR, but as a 'cultural OS' that changes employee decision-making.
==========================================================
Since it automatically handles everything from researching projects that fit me like the one above to writing the application text, I only have to choose whether to apply or not. It's super easy.
🔄 Not 'Efficiency,' but 'Recapture'
When I talk about this, people say, 'That's efficiency.'
It certainly is efficiency. But my feeling is a little different.
Rather than efficiency, this is a recapture.
Until now, I was sacrificing high-value time—Monday mornings—to labor-intensive tasks like searching for projects and thinking about application texts. I've taken that back and can now dedicate it to the creative work that should have been there in the first place—writing articles, refining concepts, and thinking up new projects.
Spot consulting itself is a pretty good job in terms of hourly rate. I've written about this secret before.
The problem was that in order to go after that 'well-paying hour,' I was paying a 'zero-value hour' every single week.
I handed over the 'tasks' to AI, but what I got back was 'time.' And for someone living independently, time is the most precious and non-reproducible capital there is.
Spot consulting side hustles have changed from an activity that ate up my Monday mornings into a source of funds for using Monday mornings creatively. This may seem like a small change, but it is quietly changing the structure of how I work after becoming independent.
🧭 The Era of Having Your Own Personalized Skills
I think many people stop at just asking ChatGPT questions.
But the true power of AI emerges when you cultivate skills that have been given your own context.
The note automated posting skill I released in the second part of this series was exactly that.
My CLAUDE.md is still being updated every week.
The skills are also increasing rapidly. Note writing workflow, YouTube script production, project monitoring, meeting minutes generation. Each one operates while sharing the context of 'Roel'.
This doesn't feel like using a convenient tool; it feels closer to building my own alter ego piece by piece.
I think you can use this if you customize it, so I'll share the skill.
Automated Project Monitoring Skill | General Template
You can use this skill regardless of your job or industry by rewriting ●●● to ●●●. Copy and paste it and give it to an AI (Claude/ChatGPT, etc.).
■ Overall Flow
STEP 0: Load resume file STEP 1: Scan 4-5 target platforms in sequence STEP 2: Blacklist exclusion gate (filter out non-starters) STEP 3: Scoring based on 4 axes x 25 points STEP 4: Whitelist bonus points STEP 5: Final 5-question gate for quality check (-15 points if NG) STEP 6: Sort by threshold: Top Pick 85+ / Candidate 75-84 / Reference 60-74 STEP 7: Generate application text and output as a single report
■ ① Resume Profile (saved as my-profile.md)
# 私の職務経歴・スキルプロファイル
## 基本情報
- 名前: ●●●
- 主戦領域: ●●●
- 副業／顧問キャリア: ●●●年目
- 現在の主軸: ●●●
## キャリア①（最新）：●●●
- 役職: ●●●
- 期間: ●●●年〜現在
- 実績の数字: ●●●
- 得意な相談テーマ: ●●●
## キャリア②（前職）：●●●
- 役職: ●●●
- 期間: ●●●年〜●●●年
- 実績の数字: ●●●
- 得意な相談テーマ: ●●●
## キャリア③（前々職）：●●●
- （同様）
## 勝ちパターン（白リスト用・3-5個）
- ●●●
- ●●●
- ●●●
## やりたくない領域（黒リスト用・3-5個）
- ●●●
- ●●●
- ●●●
■ ② Platform Definition
# 巡回プラットフォーム
## ① ●●●
- URL: ●●●
- ログイン要否: 不要 / 必要
- ページ取得方法: 直接URL / クリック遷移 / SPA
- 1ページの件数: ●件
- 何ページ見るか: ●ページ
- 報酬体系: ●●●
## ② ●●●
（同様）
## ③ ●●●
## ④ ●●●
■ ③ Scoring Rules
Scoring is based on 4 axes x 25 points = 100 points total.
▼ Definition of 4 axes
・Experience Match (25 pts): Relevance between resume and project. Core area = high score / Adjacent = medium / Unrelated = 0 ・Compensation Appropriateness (25 pts): ●●● yen/h or more = 22-25 / ●●● yen/h = 10-20 / Less than that = 0-10 ・Project Specificity (25 pts): Is the consultation content clear and can you think of an answer in 30 seconds? ・Project Appeal (25 pts): Strategic theme / Growth company / Continuity / Uniqueness
▼ Blacklist (Exclusion gate - exclude immediately without scoring if applicable)
・●●● ・●●● ・●●●
▼ Whitelist Bonus
・Matches winning pattern: +10 pts ・●●● context is explicitly stated: +5 pts ・High unit price x brainstorming type: +3 pts
▼ Final 5-question gate (conducted before generating application text - -15 points if even one is No)
Can you provide 3 or more field episodes?
Does the project want 'first-hand experience'? (NG if it's just bird's-eye market analysis)
Are you excluded by restrictions like 'Active ●●● only'?
Can you think of an 'answer that makes the other party lean in' during a 30-second simulation?
Can you write achievements with numbers in the application text?
▼ Thresholds
・85 points or more: 🥇 Top Pick (Always apply) ・75-84 points: 🥈 Candidate (Apply if time permits) ・60-74 points: 📋 Reference (Simple application text only) ・Less than 60 points: Exclude
▼ Upper Limit
Limit your top choices to 3 per week (considering the time required for follow-up after applying).
■ ④ Application Template
▼ 75 points or higher: Full application text
Always include a 'seat declaration' in the first line (stating your perspective). Example: 'I can speak as a [role] (stakeholder/mentor/designer/manager) for [topic].'
Followed by 4 blocks:
■ 関連する実務経験
[案件要件にもっとも合う経歴を2-3個。必ず数字を入れる]
■ お話しできること
[相手が知りたい論点に直接答える形で、2-3行で具体的に]
■ 直近の活動
[現在進行形のプロジェクトを1-2行]
▼ 60-74 points: Simplified application text
[One-line strength] + [One achievement with numbers] + 'I believe I can help. Please let me speak with you.'
▼ Prohibited expressions (Avoid these as they make you sound like a mass-produced template)
・ 'I would be happy to accept' ・ 'I can speak broadly about' ・ 'Regarding industry trends'
■ ⑤ Main Prompt (Instructions for the AI)
あなたは案件巡回エージェントです。
【タスク】
1. my-profile.md を読み込み、経歴と勝ちパターンを把握する
2. 4プラットフォームを順番に巡回し、案件タイトル・報酬・期日を取得する
3. 黒リストで除外ゲートを通す
4. 4軸×25点で採点する
5. 白リストボーナスを加点する
6. 最終5問ゲートで再チェック（NGなら-15点）
7. 閾値で本命/候補/参考に振り分ける
8. 本命と候補にはフル応募文、参考には簡易応募文を生成する
9. 1本のレポートにまとめて出力する
【絶対ルール】
- 案件タイトルはプラットフォームの表示を一語一句そのままコピー（要約・言い換え禁止）
- 黒リスト該当はスコアリングに進めない
- 最終5問ゲートをスキップしない
- 応募文の数字はプロファイルから取り、勝手に作らない（虚偽厳禁）
- ログイン切れ／メンテナンス時はそのプラットフォームをスキップ、レポートに明記、他は継続
- 本命は週3件以内に絞る
【出力フォーマット】
# 週次レポート — [日付]
## 今週のサマリー
- 巡回: ●/●サイト
- 総スキャン: ●件
- 本命: ●件 / 候補: ●件 / 参考: ●件
## 🥇 本命（85+）
[各案件：タイトル／スコア内訳／報酬／時間／企業／期日／おすすめ理由／応募文]
## 🥈 候補（75-84）
[同フォーマット]
## 📋 参考（60-74）
[一覧＋簡易応募文]
## プラットフォーム別サマリー
[各プラットフォームのスキャン件数・通過件数・特記事項]
## 次のアクション
[締切順・優先度順]
■ ⑥ Application Log (For learning functions)
Save as job-search-log.json. Log the jobs you were accepted for and those you weren't, and pass this to the AI each time. By instructing it to 'add +5 points to themes similar to winning_themes next time,' the accuracy will improve with each iteration.
{
"last_run": "YYYY-MM-DD",
"applied_jobs": [
{
"date": "YYYY-MM-DD",
"platform": "●●●",
"title": "●●●",
"score": 0,
"score_breakdown": {"experience": 0, "reward": 0, "specificity": 0, "appeal": 0, "bonus": 0},
"result": null
}
],
"winning_themes": [],
"losing_themes": []
}
■ 5 Customization Steps
Create my-profile.md (30 minutes)
Select 4-5 platforms to monitor (10 minutes)
Rewrite the ●●● sections of the scoring rules for yourself (20 minutes)
Run it once and check the output (5 minutes)
Tune it 2-3 times before setting it to run regularly
If you use Claude and are interested in the above, please give it a try.
💬 If you resonate with this even a little—
It would encourage me if you could 'Like,' 'Comment,' or 'Follow.' (I generally follow back)
📺 I've started a YouTube channel.These are videos you can listen to casually during your morning commute.
OS for Living Outside the Organization: Roel
Please subscribe to the channel 🙏