Knowledge for Account Assist in Outreach (Beta)

Created by Tomas Zeman, Modified on Fri, 20 Feb at 6:16 PM by Joel Sandi

Note: Feature(s) Pending Rollout

Features and functionality described on this page are part of our February 2026 release, which is rolling out to eligible customers February 12-March 12, 2026. See our Release Notes here to learn more

Objective

This article explains how Knowledge helps admins upload sales collateral into Outreach so that Account Assist can produce more relevant, consistent, and trustworthy answers by referencing customer-provided documents. It will allow our AI to learn about you.

Applies To

  • RevOps Leaders

  • Outreach Admins / System Admins

Before you begin

Before uploading documents, make sure you have:

  • Admin access to manage Knowledge documents in Outreach

  • Your sales collateral ready (examples below)

  • Documents that are clear, accurate, and up to date (avoid outdated messaging or obsolete pricing)

  • An understanding of how Playground vs Account Assist channels work (recommended workflow is: test first → then promote)

Recommended document types:

  • Company overview

  • Product positioning / feature overviews

  • Personas, ICP, and pain points

  • Value propositions and messaging frameworks

  • FAQs and objection handling

  • White papers

  • Case studies and proof points

  • Security / compliance documentation

Overview

Knowledge lets RevOps leaders and admins upload sales collateral into Outreach so that Account Assist can generate better answers grounded in approved internal content.

How it works

  1. Upload documents (PDF, Word, or text files) into Knowledge

  2. Outreach processes and indexes the document content

  3. When a user asks a question in Account Assist, the system checks for relevant documents

  4. If relevant content is found, Account Assist can reference it to form a stronger response

  5. Users can see referenced materials in the response Sources (when available)

This approach is commonly referred to as Retrieval-Augmented Generation (RAG) — meaning the AI uses your trusted content to reduce inconsistency and increase confidence in responses.

Recommended workflow

  1. Create a directory structure in Outreach for your documents.

  2. Upload documents into Outreach

  3. Add documents to the Playground channel

  4. Validate content quality safely in the Ask in Playground chat. Here you can see AI answer without and with the documents in the Playground channel. 

  5. When tested, add documents into Account Assist channel → enable approved content for real end-user workflows

Additional Information

Channels explained

Channels control where a document is allowed to be used:

  • No channel selected
    Document is stored, but not used anywhere

  • Playground
    Document is available for testing and validation in the Playground

  • Account Assist
    Document is actively used to support Account Assist responses 


Document statuses

During upload and processing, documents may display statuses such as:

  • Uploading – document is being uploaded

  • Uploaded – upload finished successfully

  • Processing – document is being parsed and indexed

  • Processed – ready for use in channels

  • Error – document could not be processed (details shown in UI)


Best practices

  • Start with high-signal documents first (FAQs, value props, personas, case studies)

  • Keep your library clean and relevant — adding too much low-quality content can reduce answer quality

  • Always validate in Playground before promoting to Account Assist

  • Prefer text-heavy documents (the system relies on extracted text)


Limitations (example formatting)

Depending on your plan and configuration, Knowledge may enforce limits such as:

  • maximum file size - 10MB

  • total storage capacity - 500MB

  • maximum number of documents - 1000

If you need higher limits, contact Outreach Support.


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