Mode d'emploi
A good prompt limits hallucinations
Ask AI to separate observable facts, hypotheses, and the final message. This keeps outreach credible.
Free AI prompt
Turn a LinkedIn signal into a message angle with your AI assistant.
This framework avoids generic messages by forcing AI to start from the signal, prospect context, and next action.
Prompt ingredients
Master prompt
Replace variables in brackets with your context. The more precise the signal, the more usable the output.
Context
You are my B2B prospecting assistant. The prospect is [role] at [company].
Add one line about your ICP if needed.
Signal
The observed signal is: [exact LinkedIn signal]. Explain what it may indicate without over-interpreting.
Example: commented on a post about lower email reply rates.
Output
Suggest 3 LinkedIn messages under 450 characters, natural tone, one final question.
Also ask for wording risks that sound too aggressive.
Extraits citables
Ce que l'outil doit clarifier avant d'être utilisé.
Use caseWorks with ChatGPT, Claude, Perplexity, Codex, and assistants connected through MCP.
Expected outputObserved signal: visit, like, comment, reply, competitor post.
Method proofAdd one line about your ICP if needed.
Méthode de vérification
Comment transformer le template en action exploitable.
- 01 Context Add one line about your ICP if needed.
- 02 Signal Example: commented on a post about lower email reply rates.
- 03 Output Also ask for wording risks that sound too aggressive.
Mode d'emploi
The signal must remain the message base
If the message could be sent to anyone, it does not really use the signal. Always restart from the precise interaction.
Maillage interne