Using Knowledge Bases with other Profound features

All three components of the Knowledge Bases section connect to Profound's Agents. Knowledge Bases and Brand Kits are available as dedicated nodes available to any Agent workflow, and Audience Segments can be applied at the content generation layer. Knowledge Bases also serve as the source of truth for FactCheck, Profound's brand information accuracy verification feature.

Knowledge Base Agent nodes

Four dedicated nodes are available in the Agent Builder: Prompt LLM, Search Knowledge Base, Get Knowledge Base File, and Upload to Knowledge Base. Each node addresses a different stage of an Agent workflow.

Prompt LLM node (with Use Knowledge Base enabled)

When Knowledge Base access is enabled within a Prompt LLM node, the model references the Knowledge Base content to generate its output.

Best used when: You want the LLM to ground the generated content in your brand's approved documentation.

Search Knowledge Base node

When you’re unsure which documents are most relevant, use this node to perform a search across your Knowledge Base. Results are ranked by relevance and passed into the following nodes for summarization, analysis, or content generation.

Best used when: The Agent needs to find the right source document before acting. For example, it might retrieve the most relevant product spec before generating a comparison article.

Get Knowledge Base File node

This node retrieves a specific file from your Knowledge Base, such as a brand style guide, product specification, or compliance document. The file is passed directly into subsequent workflow steps.

Best used when: The Agent always needs the same document at a fixed point in the workflow.

Write to Knowledge Base node

Enables Agents to upload newly generated documents directly back into your Knowledge Base in one automated motion.

In the node configuration, specify the Knowledge Base (upload location), content, and a document name. If a document with that name already exists in the Knowledge Base, its contents are overwritten. If it doesn’t, a new document is created.

Best used when: You want to reuse a valuable Agent node output (like competitive research reports or performance summaries) in future workflows.

Brand Kit Agent nodes

Brand Kits connect to Agent workflows through one dedicated retrieval node and three content generation nodes that accept Brand Kit variables as inputs.

Use Brand Kit node

The Use Brand Kit node retrieves a single field from any of your saved Brand Kits and outputs it as a named variable. Available fields cover all Brand Kit sections: About the Brand, Ideal Customer Profile, Competitors, Brand Point of View, Author Persona, Tone of Voice, and Writing Rules.

Best used when: You want to pass a specific piece of brand context, such as tone of voice or writing rules, into a content generation or LLM step.

Prompt LLM node

The Prompt LLM node can reference output variables from upstream Use Brand Kit nodes, letting you inject any Brand Kit field directly into the model's instruction prompt. This is independent of the node's Knowledge Base access option: use both in the same step, or either on its own.

Best used when: You need full control over how Brand Kit data is applied in a custom instruction. For example, pass Tone of Voice and Writing Rules as variables into a Prompt LLM node to rewrite content in the brand's established voice.

Generate Article node

The Generate Article node accepts Brand Kit output variables in its Additional Instructions field, as well as Audience Segment variables in an Audience Segments field. It also includes a built-in brand alignment pass: a final refinement stage that checks the article for brand-safe language, correct terminology, and consistent voice. Supplying Brand Kit data from upstream nodes gives this pass the information it needs to apply your specific brand standards.

Best used when: Building end-to-end automated content workflows where every generated article must reflect brand-approved tone, style, and positioning.

Create Content Brief node

Brand Kit variables from upstream Use Brand Kit nodes can be passed into the Create Content Brief node via the Additional Instructions field, carrying brand context through to any downstream Generate Article step. The node also accepts an Audience Segment field natively, so brand voice and audience targeting can inform the same brief.

Best used when: You want to produce a content brief that already incorporates brand voice guidelines, ensuring writers and downstream generation steps are working from a brief that reflects both research and brand standards.

Combining node outputs

Like all Profound data nodes, Knowledge Base nodes produce a named output variable that can be passed into any subsequent node that accepts text or structured data inputs.

For example, a Search Knowledge Base node can surface the most relevant product documentation, a Use Brand Kit node can retrieve the brand's tone of voice, a Prompt LLM node can combine both to draft an accurate, on-brand comparison article, and a CMS node can publish the result directly. The Upload to Knowledge Base node closes the loop by storing finished outputs back into your Knowledge Base for reuse in future workflows.

Don't be afraid to experiment with different node combinations. The workflows that work best for your team will depend on your content goals, publishing cadence, and the structure of your Knowledge Base.

Knowledge Bases and FactCheck

The FactCheck feature uses your Knowledge Base as its source of truth to verify the claims AI engines make about your brand. Before FactCheck can run, upload your brand documentation to a Knowledge Base: support pages, product specifications, policy documents, and Q&A files all work well. Once set up, FactCheck extracts AI claims about your brand and verifies them against your indexed content, allowing you to pinpoint and address the misinformation.