Coaches who want to streamline session notes, create powerful worksheets, and generate follow‑up emails can use Claude, Anthropic’s conversational AI. This guide walks you through setting up Claude, prompting it effectively, and integrating the output into a typical coaching workflow. Follow each step and you’ll see measurable time savings while keeping the personal touch that clients expect.
Visit anthropic.com and click “Sign Up”. Choose the “Coach” plan if available; otherwise the “Starter” plan gives 100 k tokens per month free.
Open your browser console and run a simple fetch request. Replace YOUR_KEY with the copied key.
fetch('https://api.anthropic.com/v1/complete', {
method: 'POST',
headers: {
'x-api-key': 'YOUR_KEY',
'Content-Type': 'application/json'
},
body: JSON.stringify({
model: 'claude-3-opus-20240229',
max_tokens_to_sample: 50,
prompt: '<Human> Hello Claude!</Human>\n<Assistant>'
})
}).then(r=>r.json()).then(console.log);
If you see a JSON response with a short greeting, you are ready.
Claude follows the style you set in the prompt. A good starter is:
You are a supportive life coach assistant. Summarize client statements, ask powerful questions, and suggest three actionable steps.
Claude‑3‑sonnet supports up to 100 k tokens. For a 60‑minute Zoom transcript (≈9 k words) you can send the whole text. For longer programs, send the last two sessions plus a summary.
---Begin Transcript---
[Paste raw Zoom transcript here]
---End Transcript---
Summarize the client’s main challenges in 3 bullet points. List any goals they mentioned. Suggest two follow‑up questions for next week.
Use a tool like Otter.ai or Zoom’s native transcription. Export as .txt.
Copy the transcript into the prompt format from 2.3. Run the request via a simple Python script (no external libraries needed).
import json,requests,os
API_KEY = os.getenv('CLAUDE_KEY')
TRANSCRIPT = open('session.txt').read()
prompt = f"""---Begin Transcript---
{TRANSCRIPT}
---End Transcript---
Summarize the client’s main challenges in 3 bullet points. List any goals they mentioned. Suggest two follow‑up questions for next week."""
payload = {
"model":"claude-3-sonnet-20240229",
"max_tokens_to_sample":200,
"prompt":prompt
}
resp = requests.post(
"https://api.anthropic.com/v1/complete",
headers={"x-api-key":API_KEY,"Content-Type":"application/json"},
data=json.dumps(payload)
)
print(resp.json()['completion'])
Claude’s output is a draft. Read it, correct any mis‑heard phrases, then save as a PDF note for the client.
Most coaches use three sections: Reflection, Action, Accountability.
Create a worksheet for a client who wants to improve time management. Use these sections:
1. Reflection – three prompts that make the client examine current habits.
2. Action – three concrete tasks to try this week.
3. Accountability – a simple table where the client logs start/end times.
Provide the worksheet in markdown format.
Paste Claude’s markdown into an online converter (e.g., Dillinger) or use Pandoc locally:
pandoc worksheet.md -o worksheet.pdf
Write a friendly follow‑up email for a client named Alex who just completed a goal‑setting session. Include:
- A brief recap of the three goals.
- The next‑step tasks from the worksheet.
- An invitation to schedule the next call.
function sendClaudeEmail() {
var apiKey = PropertiesService.getScriptProperties().getProperty('CLAUDE_KEY');
var prompt = 'Write a friendly follow‑up email for a client named Alex...'; // truncated
var options = {
method: 'post',
contentType: 'application/json',
headers: {'x-api-key': apiKey},
payload: JSON.stringify({
model: 'claude-3-sonnet-20240229',
max_tokens_to_sample: 250,
prompt: prompt
})
};
var response = UrlFetchApp.fetch('https://api.anthropic.com/v1/complete', options);
var emailBody = JSON.parse(response.getContentText()).completion;
MailApp.sendEmail('alex@example.com', 'Your Coaching Follow‑Up', emailBody);
}
Run the script after each session. You can schedule it with a trigger for weekly automation.
Anthropic offers a claude-3-opus-20240229:private model that does not retain data for training. Switch to it for any client‑identifiable content.
Before sending a transcript, replace names with placeholders:
Client: [CLIENT_NAME]
Coach: [COACH_NAME]
Save all Claude outputs in an encrypted folder (e.g., using VeraCrypt). Keep the decryption key in a secure password manager.
| Feature | Claude 3 Opus | ChatGPT‑4 Turbo | Google Gemini 1.5 |
|---|---|---|---|
| Max context window | 100 k tokens | 128 k tokens | 30 k tokens |
| Privacy‑only mode | Yes (private endpoint) | No | Limited |
| Cost per 1 k tokens | $0.015 (Opus) | $0.03 | $0.012 |
| Best for coaching tone | High (system prompts work well) | Medium | Low |
| Ease of integration | Simple REST, no SDK needed | SDKs available | SDKs needed |
For most coaches, Claude offers the best blend of privacy, cost, and ability to keep a supportive tone. ChatGPT‑4 provides a larger context but lacks a true private mode, which can be a compliance risk.
Claude is an Anthropic large‑language model that can generate text, summarize sessions, and draft action plans. Coaches use it to save time, keep notes consistent, and spark new questions for clients.
A free trial account works for most steps. Some advanced features, like larger context windows, require a paid plan.
No. Claude is a tool that amplifies your expertise. You still decide which insights to share and how to guide the client.
Use Claude’s “private” endpoint, avoid sending personally identifiable information, and store all transcripts on your own encrypted server.
Claude runs only on Anthropic’s cloud. You only need a modern browser and an internet connection.
By following these steps, coaches can incorporate Claude into every phase of their practice—from live session capture to post‑session follow‑up. The result is clearer communication, faster paperwork, and more time to focus on the human side of coaching.