Introduction

Hello. My name is Yamazumi and I work in sales at Classmethod.

At Classmethod, not only engineers but also sales and back-office staff are using generative AI to improve work efficiency.
We catch up on update information that appears rapidly every day and incorporate it into our work.

Among these, the number of employees using Claude has been especially increasing. Until recently, I mainly used Gemini within Google Workspace, but lately Claude has become my primary tool. Claude has significantly increased its capabilities.

Today, I'll introduce Sales Plugin, a set of functions for sales professionals that can be used on Claude.
Along with each function, I'll share my review of how it worked for me as someone engaged in frontline sales activities.

What is Claude's "Sales Plugin"?

How to start using plugins?

The Sales Plugin is Anthropic's sales support plugin that can be used in Claude desktop app's Cowork mode.

Installation is simple. You can install it using either of the following methods:

• With the Claude desktop app installed, install the plugin from the Claude official site
• From the "New Task" screen in Claude desktop app's Cowork screen, click the "+" button and go to "Plugins" > "Add plugin"
  
  ⇧ You can add it from here.

For plugin features, there are methods to add from "Anthropic & Partner" (plugins created by Anthropic official or partners like SalesForce) and methods to add ones created by general users from GitHub or marketplaces.

The Sales Plugin I'm introducing today is officially created by Anthropic.
For non-official plugins, some may include ones that haven't been verified for operation, so caution is needed when using them.

I have only used the Anthropic official ones so far.

What's inside the plugin

The "Sales Plugin" might be easier to imagine as a "collection of useful functions for sales professionals."

The plugin consists of 9 Skills.
(When the plugin was first released, it consisted of 6 Skills and 3 Commands, but after a recent plugin update, everything has been integrated into Skills.)

What are Skills
These are functions that are automatically triggered during conversations between users and Claude. For example, if you ask "research about Company X," Claude will determine "I should use that skill in this case!" and launch the appropriate skill for you. You can also explicitly call them using the / command.

You can check the list of Skills and the contents of each Skill from the customization screen. You can also edit them. You can have Claude edit them by conversing with Claude, or you can directly edit the Markdown files yourself.

⇧ You can check the contents of each Skill like this.

You can also connect with external tools like CRM (SalesForce, HubSpot, etc.), calendar, email, Slack, etc. using connectors.
Here are the connectors prepared by default in the Sales plugin.

You can manage permissions for each tool. You can set whether to give only read permissions or write permissions for each service and function. This is an example of Google Calendar.

While connecting enhances functionality significantly, it's fine to start without external service connections for now.

Skills can also be customized later. If you're unsure, I recommend trying them individually with default settings first and then customizing as needed.

Introduction to all 9 features

From here, I'll introduce all features along with my actual usage experience.

1. Account Research

Just by saying "research about X Corporation," it will retrieve the following information in one go:

• Company overview (business content, size, industry)
• Latest news (funding, M&A, management changes, etc.)
• Hiring trends (what kind of talent they're looking for, if they're expanding, etc.)
• Key person information (management, decision makers)
• Overview of products/services

Let's actually try it. I'll instruct it to "research about Classmethod Inc."

What appears at the bottom of the chat is the working folder name. I specified a folder for writing this blog. By specifying a folder, I can have files created in this session saved here.
If there is other information that should be referenced, putting it in this folder is sufficient.

Although I only gave one line of instruction, as shown in the "Progress" on the right, it makes a work plan and proceeds with the research.
I have enabled the HubSpot connector, so it includes checking HubSpot, but for now I'll have it proceed with research using only web search. You can just instruct Claude "Please use only web search, not connectors" from the side while it's working.

The research was completed in just a few minutes. I think it took less than 5 minutes.

Here's the account research report that was created:

# Research: クラスメソッド株式会社（Classmethod, Inc.）
**Generated:** 2026年3月16日 **Sources:** Web Search

---

## Quick Take

クラスメソッドは日本最大級のAWSプレミアパートナーであり、クラウド総合支援・データ活用・生成AIを中心に急成長を続けるSIer。売上高は2025年度に950億円を突破し、グローバル展開（ドイツ、カナダ、インド、韓国、マレーシア等）も加速中。直近ではAnthropic社とのリセラー契約やdbt Labs公式トレーニングパートナー認定など、AIとデータ領域でのパートナーシップ拡大が顕著。

---

## Company Profile

|Field|Value|
|---|---|
|**会社名**|クラスメソッド株式会社|
|**Website**|[https://classmethod.jp/](https://classmethod.jp/)|
|**業種**|システムインテグレータ（SIer）/ クラウドコンサルティング|
|**従業員数**|単体 560名 / グループ全体 840名（2025年10月時点）|
|**本社**|東京都港区西新橋1-1-1 日比谷フォートタワー26F|
|**設立**|2004年7月7日|
|**資本金**|1億円|
|**売上高**|950億5,600万円（2025年度）|
|**代表者**|横田 聡|
|**平均年齢**|37.5歳|

### What They Do

企業向けにAWS・Google Cloud等のクラウド総合支援、データ分析基盤構築、生成AIコンサルティング、アプリケーション開発（LINE/iOS等）、SaaS導入支援を提供。直取引で業界・規模問わず4,000社以上の支援実績を持つ。技術ブログ「DevelopersIO」は30,000本以上の記事を公開し、技術コミュニティへの貢献でも知られる。

### 売上高推移

|年度|売上高|
|---|---|
|2019年度|74億円|
|2020年度|131億円|
|2021年度|216億円|
|2022年度|304億円|
|2023年度|427億円|
|2024年度|590億円|
|2025年度|**950億円**|

→ 過去6年間で約13倍の急成長

---

## Recent News（直近90日）

- **Anthropic社とリセラー契約を締結**（2026年3月2日）— Amazon Bedrock経由でのClaude再販が可能に。エンタープライズ企業の生成AI活用を推進。
- **日本初のdbt Labs公式トレーニングパートナー認定を取得**（2026年3月10日）— データ活用基盤支援の強化。
- **マレーシア・クアラルンプールオフィス開設**（2026年1月）— グローバル展開のさらなる加速。
- **Google Cloud Partner Top Engineer 2026受賞**（2026年1月）— 5名のエンジニアが受賞。マルチクラウド対応力の証明。
- **Classmethod Showcase開催発表**（2026年2月）— データ活用を事例で学ぶオンラインカンファレンス。

---

## Hiring Signals（採用動向）

- **87件の求人**を公式キャリア採用サイトで公開（2025年1月時点）
- 主な募集領域：プロジェクトマネージャー、エンタープライズマイグレーション、Google Cloudデータ分析、セールス/カスタマーサクセス、事業企画/マーケティング
- **新卒採用**もマイナビ2027で展開中
- 成長指標：年間売上が前年度比60%増のペースで、積極採用が続いていると推察

---

## Key People

### 横田 聡 — 代表取締役社長

|Field|Detail|
|---|---|
|**役職**|代表取締役|
|**備考**|クラスメソッド創業者。AWS関連事業を日本トップクラスに育て上げた|

---

## グループ会社

|会社名|概要|
|---|---|
|**プリズマティクス株式会社**|EC/CRM基盤|
|**ネクストモード株式会社**|SaaS導入支援|
|**クラスメソッドオペレーションズ株式会社**|運用支援（旧アノテーション、2025年12月社名変更）|
|**プロパゲート株式会社**|—|
|**アンダースコア株式会社**|—|
|**クラスメソッドテクノロジーズ株式会社**|—|

---

## 受賞歴・パートナーシップ

### AWS関連

- ✅ **SI Partner of the Year – GLOBAL**（2022）— グローバル最優秀SIパートナー
- ✅ **SI Partner of the Year – APJ**（2023）— アジア太平洋最優秀
- ✅ **Education Consulting Partner of the Year – Global**（2024）
- ✅ **Consulting Partner of the Year – Japan**（2025）
- ✅ **AWS生成AIイノベーションセンター PIA**に日本企業初参画（2025年11月）
- ✅ AWS生成AIコンピテンシー取得（2025年6月）
- ✅ AWSプレミアティアサービスパートナー（2015年以降継続）
- ✅ 3,500以上のAWS認定資格保有

### その他

- ✅ LINEヤフー Partner Award 2025「Best LINEミニアプリ Award」
- ✅ Snowflake「Rising Star Partner of the Year」
- ✅ 三菱UFJ銀行DX支援スキーム参画

---

## 事業領域

|サービス|概要|
|---|---|
|**クラウド総合支援**|AWS/GCPの設計・構築・運用・移行（5,000件以上、35,000+AWSアカウント管理）|
|**データ活用基盤構築**|Snowflake、dbt、Redshift等を活用したデータ分析環境の構築|
|**生成AIコンサルティング**|Anthropic Claude、Amazon Bedrockを活用したAI導入支援|
|**アプリ開発**|LINE、iOS等のアプリケーション企画・開発|
|**SaaS導入支援**|各種SaaSの導入・活用コンサルティング|
|**EC/CRM基盤**|プリズマティクスによるEC・CRMシステム構築|

---

## 主要クライアント（公開情報）

Honda、資生堂、サントリー、カプコン、スシローなど大手企業を多数支援

---

## グローバル拠点

東京（本社）、大阪、福岡、札幌、ベルリン（ドイツ）、バンクーバー（カナダ）、バンガロール（インド）、ソウル（韓国）、クアラルンプール（マレーシア／2026年1月〜）

---

## Sources

- [クラスメソッド 会社情報](https://classmethod.jp/company/)
- [ニュース・プレスリリース](https://classmethod.jp/news/)
- [Anthropicリセラー契約プレスリリース](https://prtimes.jp/main/html/rd/p/000000441.000014901.html)
- [キャリア採用サイト](https://careers.classmethod.jp/)
- [数字でわかるクラスメソッド](https://careers.classmethod.jp/feature/number/)
- [Classmethod English Overview](https://classmethod.jp/english/company-overview/)
- [AWS GenAIIC参画ニュース](https://classmethod.jp/english/news/251114-aws-genaiic/)
- [NIKKEI COMPASS](https://www.nikkei.com/compass/company/wU1tFWw1jHpYU7A5YHQ8Lq)

Company research that I would have done by visiting corporate sites and news sites completes in minutes. Information sources are also listed with links. Even if there's information that makes you wonder "where did this come from?", you can quickly check the source.
Although I didn't specifically request it, it's nice that it picks up "news from the last 90 days." Also, it seems to be looking at external job posting sites as well. Amazing.

I think this Skill is sufficient if you just want to "get a general overview first." I use it in a way that "first see the overall picture and then dig deeper into areas of interest."

If you want a more detailed report, you can use other deep research services like Gemini's DeepResearch, or customize the Skill.

2. Call Prep

When you tell it "prepare for a meeting with Company X," it generates the following meeting preparation materials:

• Account overview and background
• Information about attendees and their interests
• Expected agenda
• Discovery question candidates
• Suggestions for conducting the meeting

Let's actually try it. Starting a session from the Cowork "New Task" screen by just entering /call-prep, you'll be asked the following. This time, let's try "enter company name."

It asks various questions. Let's go with "Discovery."

It continues to ask questions. Very thorough. When asked for the company name and participant name, I responded with "Classmethod Inc. President and CEO Satoshi Yokota."

To avoid creating something irrelevant, it confirms "what phase the meeting is in," "who you're meeting with," and "what you want to achieve" before proceeding with the work. That's helpful.

Here's the final output:

Call Prep: クラスメソッド株式会社
**ミーティング:** ディスカバリー（初回商談） **参加者:** 横田 聡（代表取締役社長） **あなたのゴール:** 課題・ニーズの把握

---

## アカウントスナップショット

|項目|内容|
|---|---|
|**会社名**|クラスメソッド株式会社|
|**設立**|2004年7月7日|
|**代表**|横田 聡|
|**業種**|クラウドインテグレーション・技術コンサルティング|
|**従業員数**|単体520名 / グループ820名（2025年7月時点）|
|**売上高**|約950億円（2025年6月期）|
|**資本金**|1億円（100%自己資本）|
|**本社**|東京都千代田区|
|**ドメイン**|classmethod.jp|
|**ステータス**|新規プロスペクト|

---

## 会社概要

クラスメソッドは日本最大級のクラウドインテグレーターで、AWS、データ分析、生成AI、モバイル開発を中心とした技術支援を事業会社向けに提供しています。「すべての人々の創造活動に貢献し続ける」を企業理念に掲げ、累計5,000社超・35,000以上のAWSアカウントを支援してきた実績があります。

主要サービスは「クラスメソッドメンバーズ」（AWS総合支援サービス、24時間365日体制）で、ストック型ビジネスモデルが売上の大部分を占めます。また、技術ブログ「DevelopersIO」（50,000記事超）による情報発信で技術コミュニティでの高い認知度を誇ります。

### グローバル展開

- ドイツ・ベルリン（2016年〜）
- カナダ
- インド
- 韓国
- **マレーシア・クアラルンプール（2026年1月 新設）**

---

## 横田 聡 社長について

### プロフィール

- **役職:** 代表取締役社長（創業者）
- **年齢:** 40代後半（2004年創業時26歳）
- **経歴:** 家電量販店店員 → ITベンチャー → 技術研修講師（月200万円の売上） → 2004年にクラスメソッドを300万円の自己資本で創業
- **特徴:** 技術者出身の経営者。VCからの出資を一切受けず、「身の丈経営」を信条とする

### 経営スタイル

- **100%自己資本経営** — 外部からの資金調達なし
- **ストック型ビジネス重視** — 大型調達や大量雇用をしない堅実な成長
- **技術コミュニティ重視** — DevelopersIOブログを経営の柱に据える
- **創業メンバーが15年以上在籍** — 「友人3人を誘って4人でスタート、今も全員在籍」

### トーキングポイント

- 「身の丈経営で売上950億円まで成長」という経営哲学への敬意
- DevelopersIOの技術コミュニティへの貢献について
- 2004年の創業から20年超、100%自己資本を貫く姿勢

---

## 最新ニュース（2025〜2026年）

### 2026年の注目トピック

|日付|ニュース|商談への示唆|
|---|---|---|
|**2026年3月**|**Anthropic社とリセラー契約を締結**（Amazon Bedrock経由の再販も可能に）|生成AI領域への本格参入。Anthropic/Claudeの活用ニーズが高い可能性|
|**2026年3月**|dbt Labs公式トレーニングパートナー認定取得（日本初）|データ分析基盤の強化に注力中|
|**2026年1月**|マレーシア・クアラルンプールにオフィス開設|グローバル展開を加速中|
|**2026年1月**|「グロースパック for LINE」機能拡充|自社サービスも積極的に展開|

### 2025年の主な実績

|実績|詳細|
|---|---|
|**AWS Partner Award**|「Consulting Partner of the Year – Japan」受賞（2025年5月）|
|**AWS Generative AI Innovation Center**|日本初のPartner Innovation Alliance参加企業に選出|
|**AI駆動開発コンソーシアム**|共同設立（取締役 大橋力丈氏が発起人）|
|**Snowflake**|「Rising Star Partner of the Year」受賞|
|**三菱UFJ銀行**|「DX支援スキーム」に参画、中堅・中小企業のDX推進を支援|

---

## 推奨アジェンダ

### 1. オープニング（5分）

- 横田社長の経歴と「身の丈経営」への敬意を示す
- Anthropic社とのリセラー契約締結など、最近の動きに触れる
- ミーティングの目的と時間配分を確認

### 2. 現状の理解（15分）

- クラスメソッドとして今、最も注力している事業領域は？
- 生成AI関連の取り組み（Anthropicリセラー契約後の展開）
- グローバル展開の方向性と課題

### 3. 課題・ニーズのヒアリング（15分）

- 現在感じている経営上の課題や、成長のボトルネック
- 技術的なチャレンジや人材面での課題
- 「こういうものがあれば」という理想像

### 4. ネクストステップ（5分）

- 次回のアクションを明確にする
- 社内の他のステークホルダーへの展開可能性を確認

---

## ディスカバリー質問

相手の状況を深く理解するための質問リスト：

### 事業全体

1. 「売上950億円規模に成長されましたが、次の成長フェーズとしてどのような領域に注力されていますか？」
2. 「AWSに加えて、Anthropicとのリセラー契約やdbt Labs認定など、パートナーシップを拡大されていますが、この戦略の背景にある考えは？」

### 課題・ペイン

3. 「グループ820名規模になられて、組織としてのチャレンジはどのようなものがありますか？」
4. 「グローバル6拠点に拡大されていますが、海外展開で特に感じている課題は？」

### 意思決定プロセス

5. 「新しいサービスやツールの導入を検討される際、どのようなプロセスで進められますか？」
6. 「技術選定において、横田社長が特に重視されるポイントは？」

### 成功の定義

7. 「今後1〜2年で『これを実現できたら大きい』と考えていることは何ですか？」

---

## 想定される反論・懸念

|懸念|対応策|
|---|---|
|「自社で技術力があるので外部は不要」|クラスメソッドの技術力を尊重した上で、補完的な価値（専門性の深さ・スピード・リソース）を提示|
|「身の丈経営なのでコストに慎重」|ROIを明確にし、ストック型ビジネスとの相性を強調。段階的な導入を提案|
|「既存パートナーとの関係がある」|競合ではなく補完関係としてのポジショニングを明確に|
|「今は忙しい・タイミングではない」|Anthropic契約やグローバル展開など、今だからこそのタイミングを示す|

---

## ミーティング後のアクション

- 商談メモを記録（`/call-summary` を実行）
- フォローアップメールのドラフト作成
- CRMにコンタクト・商談情報を更新
- 社内関係者への共有サマリーを作成

Similar to the Account Research I introduced earlier, but this output focuses on "meeting preparation." Rather than just company information, it compiles information from the perspective of "how to proceed with this meeting."
This time I assumed Yokota, Classmethod's representative, as the meeting counterpart. If you're planning to sell to our company's Yokota, please feel free to use this as a reference.

In this example, I didn't provide any information about "what kind of product I handle and what I want to propose." Therefore, the output is somewhat general and thin. However, I believe the quality of output would improve if you properly input "what products your company has and what goals you aim for in this meeting."

I haven't used this feature much yet, but while writing this blog, I started thinking "Maybe I should use this rather than Account Research?" I think it's good that it researches basic customer information and recent news based on "what to talk about in the actual meeting" rather than the former. With the current usage, I think it's useful at least for creating ice-breaking topics.

I plan to try various approaches, such as enriching the input information, to make more effective use of it.

3. Daily Briefing

When you ask "tell me today's schedule," it looks at connected services (email, calendar, CRM, etc.) and generates a briefing that organizes your meetings for the day, priority tasks, pipeline changes, and more.

In my case (with Gmail, Google Calendar, and HubSpot connectors enabled), it summarized information on the following topics:

• Today's top priorities
• Today's numbers (number of deals expected to close this month, number of meetings today, number of incomplete tasks)
• Today's meetings (what meetings are scheduled at what time, who is participating, brief summary of content. Google Calendar information)
• Pipeline noteworthy deals (list of high-priority deals scheduled to close at the end of this month)
• Email check status (whether there are emails that need replies)
• Things to do today

However, I don't struggle much with checking emails and calendars, so I don't use it very often.

Since I tend to handle many deals, I found the HubSpot pipeline management (content such as "These are the deals scheduled to close in March") useful for getting an overall view.

4. Draft Outreach

When you tell it "write an email to XX," it first researches the recipient via web search, then generates a personalized email template.

It writes text that doesn't feel templated and is tailored to the recipient's situation. For example, it might mention a new service recently released by the recipient's company while suggesting a connection to your solution.

However, since I don't struggle much with composing email text, I don't use this feature often.

I think it would be effective for salespeople who do a lot of new prospecting and outbound calls.

5. Competitive Intelligence

When you enter your company name and product, it automatically generates a competitive comparison matrix in HTML format.

The output includes:

• Feature comparison matrix
• Differentiation points
• Recent releases and trends from competitors
• Sales talk track (how to respond to specific questions)
• Countermeasures for landmine questions

Since it explicitly names other companies, I can't publish a sample in this blog, but when you enter "Classmethod Inc.," it instantly creates a decent-looking HTML comparison table.

(This is the part I can barely show. In reality, there are buttons to select other companies to the right of "Classmethod." In the image above, I've blacked these out. I hope you can see that it creates a decent-looking HTML at one go.)

It lists about 5 competitors and provides "patterns to win against Company A," "patterns to win against Company B" for each. It also provides "how to talk when a competitor's name comes up," "landmines (questions that expose the opponent's weaknesses)," and "anticipated objections and how to handle them."

Again, using it as is will result in output based on web searches, so the content tends to be general (i.e., somewhat thin).

This time I based it on Classmethod itself, but it might be interesting to also use it to research what players exist as competitors to your customer's company and what strengths each company has.

In addition to the Account-research and Call-prep introduced above, you might want to use this feature if you also want to know about competitor information.

6. Create an Asset

It generates sales assets such as landing pages, one-pagers, proposal decks, and workflow demos for customers. When you convey the customer's challenges and target, it creates custom materials that match.

It quickly creates a landing page like this.

Like the competitive battle card mentioned earlier, it creates something that looks decent at one go, but naturally, the default doesn't apply your company's logo or design regulations at all, so I don't think it's at a level that can be publicly released as is.

Also, this Skill alone can only output in HTML. While sales materials typically require PowerPoint slides, you need to install a separate Skill to handle PPTX files. This is also officially distributed by Anthropic, so it's easy to install. (Since the pptx skill is not included in the sales plugin, I won't introduce it in detail in this article.)

I think it can be used as an initial draft or for internal discussion purposes when creating materials.

7. /call-summary

Activated with the /call-summary command, when you paste in meeting notes or text, it automatically generates:

Personally, I think this is a feature that could be quite useful.

• Structured minutes (who said what)
• Extraction of action items (with assignees and deadlines)
• Draft follow-up email
• Internal summary

In addition to detailed minutes, it writes out "priority themes," "new proposal opportunities," and "action items." The minutes seem comprehensive and easy to read. Furthermore, it also creates a draft follow-up email. It's great that you can organize next actions without omissions and move on to the next action immediately.

Input information can be handwritten notes you typed in or transcription data from Gemini, etc.
(Separately, with a speech recognition model like Whisper, you can also create minutes from recorded audio data. When I tried it, it took quite a long time...
For handling audio data, the following article provides details:
https://dev.classmethod.jp/articles/claude-cowork-whisper-summary/

There are now many tools for creating minutes using generative AI, but I found it convenient to be able to do it quickly on Claude without customization.

With Gmail or CRM integration, it seems you can also send emails and register activities or tasks to your CRM directly from the Claude desktop app.
I'm not using this yet, but I encourage those who want further automation to give it a try. (※Please proceed after confirming with your system administrator in accordance with your company's policies!)

8. /pipeline-review

Activated with /pipeline-review, when you provide pipeline data (CSV upload or CRM integration), it analyzes:
(※Using CRM integration requires connection via MCP, etc.)

• Pipeline health score
• Deal prioritization
• Risk flags (stalled deals, closed date overdue, etc.)
• Weekly action plan

At Classmethod, we use HubSpot for deal management, so it reviewed based on HubSpot's "deal" information.

What I found particularly interesting was that it scores the entire pipeline on a 100-point scale, not just simple aggregation. It pointed out issues like "there's a problem with the underlying data quality. These many deals don't have amounts or close dates entered," which made me sit up straight. Sales teams, let's make sure to enter deal information properly.

It also presents high-priority deals and immediate action plans.

While it's a very useful feature, effective data analysis isn't possible unless the underlying deal information is entered correctly. Let's start by registering accurate data. (I'm reminding myself too.)

9. /forecast

Activated with /forecast, when you provide pipeline data and quota (targets), it generates:

• Weighted forecast (based on probability)
• Best/standard/worst case scenarios
• Commit vs. upside breakdown
• Gap analysis (deviation from target)

Like the pipeline-review mentioned earlier, I think this feature shows its true value when integrated with CRM.
(To reiterate, good reports can't be produced unless the underlying deal information is entered correctly...)

Whether you can master this feature largely depends on how you manage pipeline data in the first place. Rather than just how to use the tool itself, the design and correct operation of the entire business flow is important. (I'm reminding myself again.)

How to use in actual sales activities

So far, I've introduced the individual functions available in the sales plugin.
In terms of use cases within general sales activities, the image would be something like:

Before a meeting → Research the company with Account Research → For important meetings, also create preparation materials with Call Prep. If you want to know about competitor information or customer markets, research with Competitive Intelligence.

After a meeting → Organize notes & create follow-up emails with /call-summary

When competition occurs → Create battle cards with Competitive Intelligence

When prospecting → Create personalized emails with Draft Outreach

Deal management/internal reporting → Manage pipeline with /pipeline-review , /forecast

Impressions after using

I got the impression that what Claude can do has expanded further.
(Although I didn't introduce it in this article,) by installing various Skills, Claude can now handle Excel and PowerPoint tasks, and I've recently been spending more time working in the Claude desktop app.
Within that context, I found it very convenient to be able to prepare for meetings and organize next actions after meetings without going back and forth between other apps and browsers. A world where most tasks can be completed on Claude is emerging. It's also good that it's an officially prepared plugin by Anthropic, so it can be used with confidence and introduced immediately.

Among the various functions, I thought Account Research, call-prep, and call-summary were at a level that could be used immediately in practice.
For research functions, even without specific instructions, it researches corporate site information, IR materials, news articles, job postings, and more. I got the impression that it collects information much faster and more comprehensively than doing web searches myself. It also clearly shows information sources, making fact-checking easy.
For call-summary which summarizes minutes, my first impression was that the resulting minutes are easy to read. I liked that it not only lists conversation content but also summarizes topics with different levels of importance in forms like "points being emphasized," and allows for smooth transition to next actions.

In this article, I mainly introduced what can be done in the default state. The sales plugin is a collection of features that apply to many sales roles, essentially a greatest common denominator.
So, there may be cases where it "doesn't quite fit my work." In such cases, I think it's good to adjust Skills to your liking or add integrations with external services to customize it.

What I'm being careful about

While it's convenient, I think there are also things to be careful about. Here are a few things I'm mindful of:

• Always fact-check the output

Fact-checking the output is essential. Since it's AI output, it may contain incorrect information. When using AI output information in materials, I always make sure to verify the accuracy of the information provided. For information based on web searches, you can generally ask "show your information sources" and it will properly report the source, so if you have any doubts, I always check the primary information with my own eyes.

• Occasionally check that Claude isn't behaving strangely

Claude Cowork makes work plans and autonomously completes tasks for you. Depending on the content, it may work diligently for over 10 minutes. It usually proceeds without problems, but occasionally it may not progress as expected and keep retrying.

From my own experience, when I had enabled "Claude in Chrome" which allows Claude to use a web browser for operations, there was a time when it said "I can't access the connected service properly, so I'll operate the browser directly" and launched a browser to keep trying to work in an RPA-like manner. Looking at the browser operations, it seemed to be repeating ineffective trial and error, such as repeatedly showing and hiding columns in a spreadsheet, so I stopped it. When this happens, it's wasteful because the work doesn't progress but tokens are consumed, so be careful.

Also, if you suddenly give it an Excel file with a lot of data, it may waste time and tokens processing it, so I try to avoid giving unnecessary information and provide appropriate input information tailored to the purpose.

It's good to install many Skills and connector integrations, but I think it's also important to check if they're behaving as expected and if not, to reduce waste by deleting them.

• If you don't want your input data to be used for learning, opt out from the settings screen

This is not limited to the sales plugin. By default, Claude has the setting "Allow your chat and coding session data to be used to train and improve Anthropic AI models" enabled. If this is not acceptable to you, please use it after turning off this setting in the settings screen.

Lastly

The Claude Sales plugin has 9 features in total, but I don't think you need to master all of them.

For now, I recommend trying the ones that seem easy to use or suitable for your work, and then customizing Skills or adding other Skills as needed.

"Customizing Skills" may sound difficult, but it's not at all. You don't need to write complicated code. In the Cowork chat screen, when you say "please modify this Skill like this," it will run "Claude's Skill for creating Skills" and make nice adjustments for you without any problems.
After modifying, you can test it again, and if there's anything you don't like, you can make further adjustments. This cycle of gradual improvement becomes increasingly enjoyable.

For those curious about "what it's like," I recommend trying Account Research, Call-Prep, and /call-summary first. These work reasonably well even without other Skills or external connector integrations.

There are many plugins besides sales. I think it's fun to find ones that suit your own work.

Although I'm just a sales representative speaking rather presumptuously, personally I believe that generative AI is not a replacement for sales but rather a good assistant that improves the quality of sales. I want to continue exploring various ways to use it, primarily focusing on how to enhance the quality of negotiations and communications with customers.

If you think "I'm not an engineer, so Claude is irrelevant to me," especially sales people! Please try using Claude!

That's all.