How to set up knowledge base for lab results timing statu…

• A Chatref account with an active workspace. Every new account includes $50 of free credit (no card required, never expires) and unlimited AI agents. If you don’t have one yet, start at app.chatref.ai.
• Your lab’s documents covering results timing. You’ll need clear, up-to-date information on turnaround times per test type (e.g., CBC, culture, PCR), how patients can check status, your operating hours, and any special handling rules (stat vs. routine). The more precise the wording, the better the agent will match it.
• Where patients will find the chat. A patient portal, your lab’s website, or a results page where you can add the Chatref widget snippet.

If you’re new to using AI for lab patient communication, our Laboratory Services overview explains how Chatref handles routine questions, freeing your staff for clinical work. This guide walks through the specific setup for results timing and status tracking.

1. Gather your results timing and status information

Collect the exact language you want your agent to speak. Typical sources:

• Turnaround time tables (sorted by test and collection method).
• Patient-facing instructions (e.g., “Results available in your portal 24 hours after the sample is processed”).
• FAQs about delays, redraws, or after-hours availability.
• Escalation criteria for your staff (e.g., “Positive critical results are called to the ordering provider directly”).

Save these as a PDF or a set of well-formatted text files, or make sure the live pages are accessible via URL.

2. Add the content to your Chatref knowledge base

In your Chatref workspace, go to the Knowledge Base section.

• Click Add Document, then choose Upload, URL, or Paste text.
• Upload your PDF or add the URLs of your public turnaround-time pages.
• Give the document a descriptive name (e.g., “2026 CBC and Chemistry TAT”) so you can update it later.
• Repeat for each piece of content. Chatref reads everything as plain text, so avoid image-only PDFs – embed text or provide a companion document with the text.

After adding, the knowledge base processes the content in seconds. No training step is needed; the retrieval happens at query time.

3. Build (or update) your AI agent

Go to Agents and either create a new agent or open the one assigned to your lab’s public chat.

• Under Knowledge Base, link the documents you just added.
• Set the Agent prompt to focus the agent on timing and status topics. For example:

  You answer patient questions about lab results turnaround times, status checks, and follow-up steps. Use only the lab’s own documents. If you can’t find a specific timeline, tell the patient how to reach the lab directly.

• Keep the tone professional, caring, and non-diagnostic – the agent is about logistics, not medical advice.

4. Set up a custom action to collect patient context (recommended)

A bare knowledge base can’t look up an individual’s results. A custom action bridges that gap.

In Custom Actions, create a new action.

• Name: Capture patient details for status.
• Description: Tells the agent when to trigger it (e.g., “Trigger this when a patient asks about the status of their personal results or gives an order number.”)
• Fields: Add input fields for patient name (text), date of birth (text), order/accession number (text), and test name (text).
• Action URL (optional): If your lab system exposes an API for results lookup, set the endpoint. Otherwise, leave the URL blank – the action will still collect the details and pass them to the agent, which can then provide the general turnaround time from your documents and suggest next steps.

Link the custom action to your agent under the agent’s Actions tab.

5. Deploy the widget on your site

Copy the embed snippet from the agent’s Widget tab. Paste it into your patient portal’s HTML just before the closing </body> tag. Add your domain to the allowlist if Chatref prompts you to.

The chat bubble will appear immediately. On the free credit only, there’s no feature gating – unlimited agents, branding removal, and lead capture are all included.

Open the Playground for your agent. Ask a few realistic questions:

• “How long do CBC results take if I had my blood drawn this morning?”
• “What’s the status of my COVID PCR from yesterday?”
• “I need my lipid panel results – how do I check them?”

The agent should pull the timelines from your document and not guess. It should also trigger the custom action when appropriate. Test with incomplete details or a test you didn’t document: the agent should either fall back to general lab contact information or prompt for more specifics.

Walk through the custom action flow: the agent should collect the fields you defined and, if no live API is connected, respond with the relevant turnaround guidance from your knowledge base.

Visit your actual patient portal and start a chat from a real page. Confirm the widget loads and the agent answers correctly from the live knowledge base.

• Agent returns a vague or wrong timeline.
  The exact test wasn’t in your knowledge base. Add a new document or a line to an existing one with the specific turnaround (e.g., “Urine culture: preliminary at 24 hours, final at 48-72 hours”). The agent retrieves wording verbatim, so detailed test names help.

• Custom action doesn’t trigger when a patient asks about their results.
  Adjust the action’s Trigger description to match more of the phrasing patients use. For example, add examples like “my result”, “order number 123”, or “where are my labs”. Test in the Playground after editing.

• Widget doesn’t appear on the page.
  Check that the snippet is placed on all pages where you want the chat, that the domain is allowed in the agent’s widget settings, and that no ad-blocker or CSP policy blocks the Chatref script.

• Agent gives generic internet answers, not your content.
  Verify the agent’s knowledge base is linked and that it has documents attached. If the documents were added after the agent started, the agent may need a quick restart – toggle the agent off and on in the settings to ensure the latest content is loaded.

What causes lab results timing status chat problems for Laboratory Services?

Most issues stem from missing or ambiguous turnaround-time documentation. If a specific test isn’t listed, the agent can’t retrieve a timeline, leading to generic or evasive answers. Another common cause is the custom action not collecting the right patient details, so the agent can’t provide a personalized response even when the timeline is known. Finally, outdated documents (e.g., pre-2026 TATs that changed) will produce wrong answers.

How do I improve lab results timing status chat for Laboratory Services?

Keep your knowledge base tightly focused on results logistics. Update turnaround-time tables as they change – Chatref doesn’t learn automatically; you must re-upload or update the source document. Expand your custom action fields to include the most common identifiers your patients use (order number, DOB, test name). Monitor the Conversation Inbox to spot questions the agent failed on, then add that exact information to the knowledge base. For labs with a client-facing API, connect the custom action URL to return live status – that upgrades the experience from “general timeline” to “your specific result is in progress”.