AI Knowledge Bases

Knowledge Bases let you organise the content your AI agents draw on when answering questions, handling calls, and running workflows. Instead of a flat list of entries, you group related content into named collections and attach those collections to specific agents. Every agent grounded in a knowledge base gives answers from what is in it.

Manage all your knowledge bases at https://app.trustpager.com/ai-agents/knowledgebase.

Knowledge Bases vs Agent Memory

These are two separate systems and it is worth understanding how they differ before you start.

• Knowledge Bases — content you curate and upload. FAQs, SOPs, pricing pages, policy documents, product details. You control what goes in; agents read from it.
• Agent Memory — what agents automatically learn as they work. When an agent sends a follow-up, handles an enquiry, or reviews a deal, it can write a memory about that contact or opportunity so the next run picks up where the last left off. You do not feed memories manually; agents write them. Browse them at https://app.trustpager.com/ai-agents/memory.

Use Knowledge Bases for reference material you want every relevant agent to know. Use Memory to track what agents have learned through experience.

Creating a knowledge base

1. Go to https://app.trustpager.com/ai-agents/knowledgebase.
2. Click New knowledge base and give it a name that reflects its purpose — for example, "Support FAQs", "Product Pricing", or "Onboarding SOPs".
3. Add sources (see below).
4. Attach it to one or more agents from the knowledge base detail page, or from inside each agent's builder under the Knowledge section.

Adding sources

A knowledge base pulls its entries from one or more sources. You can mix source types freely within a single knowledge base.

Help Center articles

Pull from your own published help content. Either select specific articles by name, or choose Sync all to bring every published Help Center article into the knowledge base as entries. Re-syncing updates any articles that have changed since the last sync and skips the ones that have not.

Product catalog

Imports your product and service listings directly from your catalog. Keeps agents accurate on pricing and availability without manual copy-paste.

Web page URLs

Paste a URL and TrustPager fetches the page and extracts its text as a knowledge entry. Useful for public documentation, terms pages, or your own website content. Re-syncing a URL re-fetches the page and updates the entry if the content has changed.

Uploaded files

Upload .txt, .md, .csv, or .html files directly. For PDF or Word documents, copy the text and paste it as a manual entry instead — those formats are not ingested directly.

Attaching a knowledge base to an agent

A knowledge base only grounds an agent if it is attached. You can do this from two places:

• From the knowledge base detail page — use the Agents section to select which voice agents and workflow agents should use this knowledge base.
• From the agent's builder — open the Knowledge section inside the voice agent or workflow agent builder and select the knowledge bases you want it to draw from.

Both directions produce the same link. Many agents can share one knowledge base; one agent can be attached to several knowledge bases at once.

Keeping knowledge bases current with resync

Each knowledge base has its own resync schedule. Choose from:

• Manual — only resyncs when you click the sync button yourself.
• Daily — re-pulls all sources once a day automatically.
• Weekly — re-pulls all sources once a week automatically.

A resync checks each source for changes, updates entries where the content has changed, and skips entries that are already current. Nothing is deleted unless you remove the source.

Adopting an existing knowledge base

If a voice agent or workflow agent was already connected to a knowledge base before the unified Knowledge Bases system launched, that knowledge base will show an Adopt option on the Knowledge Bases page. Adopting it does three things:

• Links the knowledge base into the system in place — no duplicate is created.
• Imports any recoverable web-page sources as searchable entries on the Knowledge Bases page. Sources that were text-only and whose original text cannot be recovered are not imported.
• Auto-attaches the agents that were already using it, so there is no disruption to those agents.

After adoption the knowledge base appears and is managed the same way as any other knowledge base.

Ungrouped entries

Entries that exist in your workspace but are not part of any knowledge base appear in an Ungrouped bucket on the Knowledge Bases page. They are still searchable workspace-wide but are not scoped to any specific agent. Move them into a knowledge base to make them available for agent grounding.

Permissions

Viewing knowledge bases requires knowledge:read. Creating, editing, and syncing requires knowledge:write. Manage roles at https://app.trustpager.com/settings/permissions.

Troubleshooting

• An agent is not using the content I added. Check that the knowledge base is attached to that agent. Open the knowledge base and confirm the agent appears in the Agents section, or open the agent's builder and check the Knowledge section.
• A URL source is returning outdated content. Trigger a manual resync on that knowledge base. If the page content has changed since the last sync, the entry will update.
• I uploaded a PDF and nothing appeared. PDF and Word files are not ingested directly. Copy the text from your document and paste it as a manual entry instead.
• The New knowledge base button is not visible. Your role has read-only knowledge access. Ask a workspace admin to grant knowledge:write at https://app.trustpager.com/settings/permissions.