Create and Use Personalization Agents in Outreach

Created by Anirudh Saraswat, Modified on Thu, 9 Jul at 11:42 AM by Aye Myat

Note: Feature(s) Pending Rollout

Features and functionality described on this page are part of our July 2026 release, which is rolling out to eligible customers July 9-July 28, 2026. See our Release Notes here to learn more.

Objective

Learn how to guide AI in Outreach to personalize messages with the right tone, length, context, and data sources.

AI Personalization in Outreach helps you scale relevant, high-quality messaging by automatically tailoring emails, LinkedIn messages, and call talking points for each prospect.

For best practices on how to guide AI to generate messaging that aligns with your go-to-market strategy, click here.

Applies To

  • Outreach Admins

  • Outreach Content Creators

  • Outreach Amplify Package

Procedures

Create a New Personalization Prompt

  1. Log in to Outreach as an Admin.
  2. From the left nave, click Personalization Agents.
  3. Select New personalization.
  4. Complete each of the following sections:
    1. Personalization name – Assign a name that makes the goal of this prompt and the scenario(s) where you are most likely to use it clear. This name is internal only and will not be used by the AI Agent.
    2. Gen AI instruction – Provide specific prompt instructions, following the best practices outlined here.
      • Select desired reading length and tone
      • Buyer Data– Select any relevant third-party signals (if applicable) or additional prospect or account information that you may want to include as further detail for AI to consider when drafting this message.
        • Note: The selected data points follow a waterfall method, so if Outreach does not pick up on the first signal you pick, it will default to the next, and the next, etc. We recommend you start with rarer signals like IPO announcement on the top and more common ones towards the bottom like new product launch. You can also select account and prospect information as data points—however, any information that should be included as part of the prompt no matter what should be included directly within the Gen AI instructions through variables instead.
    3. Subject line personalization– Below the Gen AI instruction, optionally toggle this setting on to have the same instruction generate a personalized subject line alongside the email body. You do not need to add separate prompt instructions for the subject line — the toggle changes what the existing Gen AI instruction outputs.
      • This is available anywhere AI Personalization is used today, including the Personalization Agent tab, Personalization added to a sequence, and single emails sent with Assist.
      • Limitation: When Subject Line Personalization is on, only one Personalization snippet can be used per email.
      • Note: Existing personalizations will continue to work as before, generating only the email body snippet. If needed, they can be edited to also generate a personalized subject line.
    4. Knowledge – Optionally, enable Knowledge to allow the AI to reference uploaded documents when generating personalised messages.
      • When turned on, you can select up to seven documents for the AI to reference. The AI will only draw from these documents when your prompt explicitly instructs it to do so.
      • If the AI references any documents, citations will be reflected in the output, and users will be able to download those documents directly.
        • Example prompts:
          • "Use the pain points and value proposition from the document context to craft a tailored message for this target persona."
          • "Use the product value proposition from the document context to highlight the key benefits of our new product XYZ."
          • "Using the {{title}} messaging from the document, write a two-sentence opening that speaks to their priorities."
      • Use cases include:
        • Reference the latest product launches and updates to keep messaging relevant
        • Apply competitive positioning to differentiate against rivals
        • Generate role-specific and industry-targeted messaging
        • Leverage NPS feedback to craft personalized follow-up outreach
        • Drive renewals by incorporating usage playbooks into personalized sequences
      • Limitations:
        • Knowledge and Seller Content cannot be enabled at the same time, enabling one will disable the other.
        • A maximum of seven documents can be selected.
        • AI Personalization with Knowledge can only be enabled when at least one document has been uploaded, processed, and enabled for the "AI Personalization" channel in the Documents section of Knowledge.
        • Document format affects retrieval quality, While PPT, PDF, DOCX, TXT, and MD formats are all supported, structure and document quality matter more than file type. Keep the following in mind when preparing documents:
          • Markdown (MD) produces the best results, clean structure, minimal noise, and accurate chunking.
          • DOCX performs well with good semantic coherence and low noise.
          • PDF offers stable content flow but may contain repeated headers, footers, and formatting artifacts that can affect retrieval quality.
          • PPTX can retain rich context via speaker notes, but inconsistent layouts in real-world decks can cause content to be extracted incorrectly, leading to unreliable retrieval
          • TXT introduces the most noise and fragmented context and is least recommended for reliable retrieval.
          • Where possible, convert content into a clean, structured format, preferably Markdown, before uploading.
      • Note: For full details on how to get the most out of Knowledge in Personalisation, see the Best Practices for AI Personalization with Knowledge article.

    5. Seller Content– Optionally, determine whether persona, industry, or product-based messaging should be used to help guide the AI’s tone and value focus to ensure the outcome is on-brand.
      • Note: More than one seller content category can be selected if appropriate.
      • If Seller Content is selected, the agent will evaluate certain prospect and account fields (as outlined below) to identify the right messaging to use. If matching content is not found, it will defer to your “Default” seller content.
        • Persona – Matches the prospect’s Outreach persona to the relevant persona-tagged content.
        • Industry – Matches the account’s industry field to the relevant industry-tagged content.
        • Product – Uses the specific product you selected to pull in the appropriate messaging.
    6. Click Save.
    7. Proceed below to utilize the new personalization prompt.

Add Personalization Prompt to Template(s)

Add your newly created personalization prompt to the relevant template(s).

Note: Personalization can be added to email templates (standalone or in Sequence), LinkedIn tasks (messages), and call tasks.

  1. In Outreach, navigate to the desired template.
  2. In the upper right corner, click Personalization.
    • Choose to either:
      • Create new personalization directly within this screen, or
      • Insert existing personalization from your library


Note: The Personalization Agent setup screen has been updated to give Admins and RevOps a clearer view of what is enabled during configuration. We recommend reviewing all sections before saving to ensure the intended combination of content sources is active and aligned with your go-to-market strategy.

Additional Information

Outreach Best Practices in Creating Personalization Agents


Was this article helpful?

That’s Great!

Thank you for your feedback

Sorry! We couldn't be helpful

Thank you for your feedback

Let us know how can we improve this article!

Select at least one of the reasons
CAPTCHA verification is required.

Feedback sent

We appreciate your effort and will try to fix the article