Skip to main content
Feedback

Quickstart: Ingest and query your first data source

note

Knowledge Hub is an Early Access release. Refer to the Early Access disclaimer for details.

This quickstart walks you through the eight steps to bring data from an Amazon S3 bucket into a working Knowledge Base and connect it to an AI agent for querying.

The example scenario ingests a folder of product documentation .docx files from an S3 bucket named acme-support-docs, so a support agent can answer product questions grounded in that documentation. Substitute your own bucket and file names throughout.

Before you begin, confirm that you have completed all items in Prerequisites, including an active Knowledge Hub role and valid AWS credentials for the S3 bucket you intend to use. If you have not yet generated a platform token or set up your Data Integration connections, complete Getting started with KH first.

Step 1: Creating a repository for your support documentation

A repository is the top-level container for your Knowledge Bases. In this scenario, the repository holds every Knowledge Base that your support organization builds from product documentation.

  1. Navigate to https://platform.boomi.com/KnowledgeHub. The Repositories page opens by default.
  2. Select Create Repository.
  3. In the Create Repository dialog, enter a Repository Title (maximum 150 characters). For example, Support Documentation.
  4. Enter a Description (maximum 1000 characters). The description is required. AI agents use it to identify and select the correct repository for a task. For example, Product support and troubleshooting content for AI agents.
  5. Select Create Repository.

Knowledge Hub creates the repository, which appears in the repositories list.

Step 2: Creating a Knowledge Base for product support content

A Knowledge Base is the queryable index within your repository. This Knowledge Base holds the product documentation that your support agent retrieves from.

  1. Select the repository you created in Step 1, then select + Knowledge Base.
  2. In the Create Knowledge Base dialog, enter a Knowledge Base Title (maximum 150 characters). For example, Product Support KB.
  3. Enter a Description (maximum 1000 characters). The description is required. AI agents use it to identify and select the correct Knowledge Base for a task. For example, Product documentation and troubleshooting guides for the support agent.
  4. Select Create Knowledge Base.

Knowledge Hub creates the Knowledge Base in Draft state and assigns the embedding model automatically. The Knowledge Base is not queryable until you publish it in Step 6.

Step 3: Connecting your S3 bucket of product documentation

Your product documentation lives in an S3 bucket, for example acme-support-docs, as a folder of .docx files. You configure the actual data connection in Data Integration, but you launch that flow from Knowledge Hub so the Repository and Knowledge Base are already selected as the target.

  1. In your Knowledge Base, select the Sources tab.
  2. Select Boomi Data Integration as the Data Provider.
  3. Select Create Source. Data Integration opens in a new tab with a source-to-target flow already targeted at this Repository and Knowledge Base.
  4. Select Amazon S3 as the source connector, then enter the credentials and bucket name (acme-support-docs) for your documentation bucket.
  5. Configure your schema to select the .docx files to ingest, and configure data preparation and chunking settings for the extracted document text.
  6. Set the ingestion schedule, or leave it as manual for this quickstart.
  7. Select Deploy.
  8. Return to the Sources tab in Knowledge Hub and select Refresh.

The deployed Source appears on the Sources tab in Draft state. Its Source Type is S3.

Step 4: Enabling the source to start ingesting your documentation

Enabling the Source starts the first ingestion run, which reads the .docx files from acme-support-docs and indexes them into your Knowledge Base.

  1. Return to Knowledge Hub, select the repository, then select your Knowledge Base.
  2. Select the Sources tab.
  3. Find the Source that was created by your Data Integration deployment.
  4. Select Enable.

The first ingestion run begins. Depending on the number and size of the files in your bucket, this can take a few seconds to several minutes. The Source transitions to Enabled state when the run completes.

To monitor progress, stay on the Sources tab and check the Last Ingestion and Error Summary columns for this Source. Select Refresh Table to update the status.

Step 5: Reviewing the schema populated from your documentation files

The Schema tab holds the canonical schema for the Knowledge Base: the full set of attributes available across all connected Sources. Once your S3 Source finishes its first ingestion run, Knowledge Hub adds that Source's mapped fields to this canonical schema automatically.

  1. In your Knowledge Base, select the Schema tab.
  2. Select Refresh Table if the attributes from your S3 Source do not appear yet. For a .docx file Source, expect attributes such as source, title, file_name, last_modified, storage_type, bucket, and key, each with a Data Type of string.
  3. Optionally, select + Attribute to add an attribute manually. In the Add Knowledge Base Schema Attribute dialog, enter a Target Attribute Name, a Description (maximum 1000 characters), and select a Data Type, then select Add.

For this quickstart, the attributes populated automatically from your .docx files are enough to test retrieval in Step 7. You do not need to add attributes manually.

Step 6: Publishing the Knowledge Base

Publishing transitions the Knowledge Base from Draft to Enabled, making it queryable by your support agent.

  1. In your Knowledge Base, select Publish in the top-right corner.
  2. Confirm the action in the dialog.
note

At least one Source must be in Enabled state before you can publish a Knowledge Base. If a precondition is missing, the General tab displays a reminder, and publishing returns a validation error until you resolve it. If this happens, confirm that your Source finished ingesting successfully in Step 4.

The Knowledge Base is now in Enabled state and is ready to serve retrieval requests.

Step 7: Testing retrieval against your documentation

Before connecting the Knowledge Base to your support agent, confirm that ingestion completed successfully and that retrieval returns relevant chunks from your .docx files.

  1. In your Knowledge Base, select the Retrieval tab.
  2. Confirm that the chunk count is greater than zero. A zero count means ingestion did not complete or no content was indexed from acme-support-docs.
  3. If the chunk count is zero, return to the Sources tab and check the Error Summary column for this Source. Common causes are invalid AWS credentials, an empty or misnamed S3 bucket, or a schema mapping error.

Step 8: Connecting an AgentStudio agent to query the Knowledge Base

Knowledge Hub does not generate responses. It returns matching chunks from your .docx files, and the agent is responsible for synthesizing them into a response.

  1. Refer to Access and permissions overview to authorize your support agent to query this Knowledge Base.
  2. In AgentStudio, add this Knowledge Base as a knowledge source for your support agent. Refer to Connecting Knowledge Bases for more information.

Your support agent can now retrieve grounded, cited answers from the product documentation in acme-support-docs.

On this Page