How to use a sales playbook for AI messages

Turn your manual sales script into a structured playbook that trains AI to write LinkedIn messages in your voice, handle objections and hit your goals.

Get started

Short answer

To train AI to write LinkedIn messages using your script, convert your manual sales process into a structured playbook. Provide specific goals for each message, the exact objections you handle, and the phrasing you use to qualify a lead.

Audit your current manual script

Most sales scripts fail because they are too broad. If your current script says you can help a company grow, the AI will repeat that vague promise. You need to find the specific triggers that make a prospect stop scrolling.

Look at your sent folder and find the messages that got the most replies. Identify the exact phrase or observation that worked. This is your hook. If you mentioned a specific job posting or a recent company announcement, that is the pattern the AI needs to follow.

List the walls you hit. Every niche has the same three or four objections. If you don't tell the AI how you personally handle the too expensive or we already have a tool objection, it will use generic corporate responses that kill the conversation.

Write down the questions prospects ask before they agree to a call. Those questions are the spine of your sequence. If every qualified lead asks who else you work with and how long setup takes, the AI should answer both before it asks for time.

Generic scripts

  • I saw your profile and think we can help you grow your revenue.
  • I would love to connect and share how our solution can help you.
  • We offer a cutting edge way to streamline your operations.

Effective scripts

  • I noticed you are hiring for three new SDRs, which usually means scaling your outbound is the priority right now.
  • I saw your comment on the post about remote work culture and wanted to ask how you're handling X.
  • Most agencies with 40 people struggle with Y. We fixed that for a similar team by doing Z.

Structure your playbook for AI

An AI doesn't need a narrative; it needs a framework. If you paste a long document, the AI will pick random parts to emphasize. Organize your script into sections that the AI can reference as rules.

The goal is the most important part. A common mistake is making every message about booking a meeting. Your playbook should define a different goal for each touchpoint. The first message should be about getting a reply or confirming a pain point, not asking for 15 minutes of their time.

  1. 1

    Define the goal for every message

    Assign a specific outcome to each step. Message 1: Confirm the prospect cares about X. Message 2: Provide a specific example of a result. Message 3: Ask for a low-friction call.

  2. 2

    Create a tone guide with constraints

    Avoid words like professional or friendly. Use constraints instead. Tell the AI to use short sentences, no corporate jargon, and never use more than one exclamation mark per message.

  3. 3

    Build a knowledge base

    List your unique value proposition in one sentence. Provide a list of the specific problems you solve so the AI doesn't invent features you don't have.

  4. 4

    Map out objection handling

    Create a list of common objections and your exact preferred answer for each. This ensures the AI stays on brand when a prospect pushes back.

Goal

Part of the playbook
The one outcome this message should produce
What to write
Get a reply that names their biggest bottleneck

Tone

Part of the playbook
Hard rules, not adjectives
What to write
Sentences under 15 words. No exclamation marks.

Proof

Part of the playbook
One result you can defend out loud
What to write
A 40-person agency cut proposal time from three days to one

Objections

Part of the playbook
The exact words you use back
What to write
We already have a tool: ask what it misses

Write the prompt for your message sequence

The prompt is the bridge between your playbook and the final message. Tell the AI exactly how to use the information you provided. If you don't specify the order of operations, the AI might jump to the pitch too early.

Keep the playbook and the prompt separate. The playbook is the reference document you reuse across campaigns. The prompt is the short instruction that tells the AI which part of the playbook to use for this sequence and which lead detail to open with.

Direct the AI to prioritize the lead's actual LinkedIn activity over the script. The script provides the logic and the goal, but the lead's profile provides the context. The best messages feel like they were written by a human who spent five minutes researching the person.

Paste this job

Act as a B2B sales expert. Use the provided sales playbook to write a 3-message sequence.

Rules:
1. Open the first message by referencing a specific detail from the lead's recent LinkedIn activity or profile.
2. Do not pitch the product in the first message. The goal is to ask a qualifying question based on the playbook's 'Pain Points' section.
3. Keep all messages under 100 words.
4. Use a low-friction call to action (e.g., 'Open to a quick chat?' instead of 'Can we book a 30 minute demo on Tuesday?').
5. Follow the tone constraints: no jargon, plain English, short sentences.

Refine the output through iteration

Your first set of AI messages will likely sound too robotic or overly eager. The fix is simple: give the AI a specific correction and a better example.

When you see a phrase that feels like AI, don't just say it sounds robotic. Tell the AI exactly why it is wrong. If it uses a word like synergy or holistic, tell it to delete that word and replace it with a plain verb. The more examples of your own writing you provide, the faster the AI will mimic your actual voice.

Example

  1. Suggested reply

    I would be honored to discuss a potential synergy between our two organizations to elevate your growth.

  2. You

    Too formal. Change to: I'm curious if you're dealing with the same scaling issues as other 40-person agencies.

  3. Suggested reply

    Our cutting-edge solution streamlines your workflow and provides robust analytics for your team.

  4. You

    Avoid adjectives like cutting-edge and robust. Say: We help you do X in half the time.

Never use the word help in your first sentence. It is a trigger word that makes prospects think you are selling something before they even finish the line.

Doing this with Kinlead

Kinlead lets you apply this playbook logic to your LinkedIn outreach without manual prompting for every lead. You can upload your sales script into the optional sales playbook field within a campaign, which the AI uses to generate personalized messages.

Because the AI handles the writing based on your goal and tone, you can focus on lead quality. You can review and edit every queued message on the Actions page before they send, ensuring the AI followed your playbook.

  1. 1

    Set up your playbook

    Paste your structured script, tone constraints, and goals into the sales playbook field in your Kinlead campaign.

  2. 2

    Define campaign parameters

    Choose your campaign goal and tone to align with the instructions in your playbook.

  3. 3

    Generate and review

    Let the AI draft the connection notes and follow-ups. Check the Actions page to edit any messages that don't hit the mark.

  4. 4

    Manage replies

    Choose whether the AI continues the conversation using the playbook or hands the conversation to you at the first reply.

Frequently asked questions

Related

Get started with Kinlead

Prefer markdown? How to use a sales playbook for AI messages.md