Train your AI Stand-in like a new teammate.

Prompts should teach judgment: audience, tone, claims to avoid, handoff rules, contact-capture timing, and what a successful conversation looks like.

Knowledge bases should carry longer reference material: docs, policies, product catalogs, implementation guides, comparison notes, and content that changes too often to paste into every prompt.

A knowledge base holds reference material that Stand retrieves when answering a relevant question. Adding a source does not retrain the model, and it does not make every answer correct. Knowledge bases are visible across your organization and used for answers by the Stand-ins to which you attach them. Every plan supports website sources; Pro and Business also support documents and sitemaps. Check Pricing for current allowances.

Use them for support manuals, docs, policies, implementation notes, product catalogs, or website sections that are too long or too changeable for a prompt. Do not use them to bury behavior instructions. The Stand-in still needs a clear prompt that explains when and how to use the source. Members can inspect shared bases and attach them to their own Stand-ins; only the member who owns a base can change or refresh its sources. Ask that person to make source corrections.

An uploaded document is a snapshot of the file you supplied. Editing your local original does not update Stand. Upload the revised file, inspect its extracted text, and delete the superseded source to avoid retrieving conflicting versions. If your allowance requires deleting the old source first, keep your original file and expect a gap until the new upload is ready.

If every Stand-in needs the same company description, tone, privacy boundary, or handoff rule, it belongs in the team prompt. If the instruction only applies to one domain or audience, put it in the site prompt. If it defines a specific Stand-in job, put it in the Stand-in prompt.

The organization owner edits Team → Team AI prompt and Sites → Add site AI prompt or Edit site AI prompt. Members edit their own Stand-in through Stand-ins → Edit → Identity. Stand combines team, site, and Stand-in instructions in that order; review all three together for contradictions rather than treating a later instruction as an absolute override.

When your positioning changes, update the shared instruction once and save it. Restart Try it out before retesting the affected Stand-ins; an existing conversation may still use the instructions it started with. Test a new website chat to check the Site layer. A prompt builder can help draft instructions, but it does not read linked pages or attached files and does not enable skills for you.

Save your changes before opening the Stand-in’s Try it out tab. If the tab is already open, choose Restart to begin a fresh test. Ask a pricing edge case, a skeptical comparison, a vague "is this for me?" question, and a question it should not answer confidently. For each test, write down the source fact or behavior you expect before reading the answer.

Try it out does not use the chat allowance, appear in the active chat list, or save a transcript in History. Save your test notes separately. A general Stand-in test does not include a matched Site AI prompt; test the published site to check that layer. Attached integration tools can make real API calls in Try it out, so use a test endpoint before exercising actions that change data.

If an answer is wrong, first check whether the needed source was indexed and attached. Then check the instructions and skill settings. Change one cause at a time, start a fresh test, and repeat the question. Finish with a real website chat: Try it out does not prove that the snippet, page rules, or live routing work.