Give your AI agent a knowledge base

Your AI agent can answer questions using your own content: FAQs, refund policies, product guides, or internal playbooks. You upload files to the agent; Chatlane indexes them so the agent can search and cite them when replying. You can also let the agent "see" images and PDFs that customers send in messages. This article explains how to set that up and why it helps your team.

Why give an agent files?

Without your own content, the agent relies on its general knowledge. With a knowledge base (uploaded files), the agent looks up relevant parts of your content and uses them to write a reply. That way, answers stay accurate and on-brand instead of generic. Support teams get consistent answers; managers can rely on the agent to follow company policies and product details.

Turning on file search

When you create or edit an agent, you can enable file search (or "knowledge base," depending on the UI). Once it's on, you can add files to that agent. The system prepares them in the background; when they're ready, the agent uses them automatically in conversations. You don't need to do anything else—just enable it and add the files you want the agent to use.

Adding and managing files

In the agent's Files section, upload documents—for example PDFs or text files. You can add or remove files anytime. After you add a file, it's processed (this may take a short time); when it shows as ready, the agent can use it in replies. If you delete a file, the agent stops using it. No technical steps—just upload and wait for the ready state.

Good candidates for uploads:

• FAQs — Common questions and answers so the agent can give accurate, consistent answers.
• Policies — Refund policy, shipping policy, terms of service so the agent can cite the right rules.
• Product guides — How-to content, feature descriptions, troubleshooting so the agent can help with product questions.
• Internal playbooks — Escalation rules, tone guidelines, or process notes so the agent stays aligned with how your team works.

How the agent uses your files

When a customer (or your team) asks a question, the agent looks up relevant parts of your uploaded content and uses them to write a reply. You don't see the search step; you just see better, grounded answers. The agent can cite or summarise your content so replies stay accurate and on-brand. Useful for support teams who want consistent answers and for managers who want agents to follow company policies.

Using attachments from messages

You can allow the agent to "see" what customers send: images (e.g. screenshots, photos) and PDFs attached to messages. When this option is on, the agent can describe or summarise those attachments in its reply. Handy for support where customers send error screenshots or documents—the agent can reference what it "sees" in the attachment when answering.

There are size and type limits for attachments (e.g. image and PDF size); the UI or help will indicate what's supported. If an attachment is too large or in an unsupported format, the agent won't use it; you can still reply manually.

Self-learning

Agents can also learn from your past conversations and build an internal FAQ that gets added to their knowledge base. That way, the agent improves over time from real support threads—without you having to write every FAQ by hand. For how to enable and use this, see Let your AI agent learn from your best conversations.

You can use both uploaded files and self-learning: upload your core content (policies, product guides) and let self-learning add knowledge from live conversations.

For step-by-step setup of your first inbox and agent, see Get started with Chatlane. For giving the agent actions (e.g. update status, tags, or send data to your CRM), see Let your AI agent take action.