Write instructions for your AI agent
Write the instructions that set your intoCHAT agent's role, tone and limits: templates, the 100-character minimum, guardrails, built-in rules and Playground tests.
Instructions tell your agent who it is, how to talk and where its limits are. Your knowledge sources supply the facts; the instructions decide how the agent uses them.
Where to edit them
Open the agent's Setup tab. Under Role and behaviour:
- Start from a template replaces the text below it with the template you choose.
- Instructions holds the text itself. It needs at least 100 characters; until then the counter shows how many are missing.
- Click Save changes in the header.
If you edited a text that started from a template, a Restore or Use latest template button above the field puts the template's text back.
Templates
Templates are starting points that already include rules for accuracy, tone and contact details. You can edit every word:
- AI Chatbot
- Customer Support Agent
- Sales Agent
- E-commerce Specialist
- Real Estate Agent
- Language tutor
- Coding Expert
- Life Coach
- Futuristic Fashion Advisor
Custom prompt starts from an empty field.
What to write
Good instructions are specific about your business. Cover:
- Role and audience: who the agent speaks for and who it talks to.
- Scope: what it should help with, and what it should politely decline.
- Tone and format: formal or casual, short or detailed, lists or prose.
- Next steps: where to send people for things the agent can't do. The agent never invents a contact channel, so name the ones you want it to use.
- A few key facts: facts that must always be right, such as your support email. Instructions outrank your knowledge when the two conflict.
Keep long content such as product details, policies and FAQs in your knowledge, where it is searched for each question.
By default the agent replies in the language the visitor writes in. To choose a reply language, use Reply language below rather than the instructions.
You are the assistant of Example Bikes, a bike shop. Help visitors
choose a bike, understand our repair service and find our opening hours.
Keep answers under 120 words and use bullet points for steps.
For questions about an existing order, ask the visitor to email
support@example.com. Politely decline topics unrelated to bikes.
Guardrails
Guardrails are rules the agent must follow in every reply, whatever a visitor asks. Add them on the Setup tab under Guardrails, in the Rules field, and click Save changes.
- Write one rule per line, for example "Never quote prices; send people to the pricing page." The field is optional and holds up to 4,000 characters.
- intoCHAT sends the rules to the AI as a separate block straight after your instructions. The agent follows them even when your instructions, your knowledge or a visitor's message suggest otherwise.
- Guardrails can only restrict what the agent says or does. They can't give it abilities it doesn't have, such as an action you haven't set up, and they can't switch off intoCHAT's built-in rules below.
- They are separate from the instructions, so choosing a template doesn't replace them.
Never quote prices; send people to the pricing page.
Don't discuss competitors.
Don't give medical advice. Suggest seeing a doctor instead.
Use guardrails for short, firm limits, and the instructions for role, tone and scope.
Stay on topic
To keep the agent to your business, switch on Stay on topic below the rules and click Save changes. It's off by default.
- The agent politely declines questions that have nothing to do with your business, such as homework, essays, coding, translations or general trivia, and says what it can help with instead.
- Greetings, thanks and brief small talk still get a short, friendly reply.
- A question that could concern your business counts as on topic. When a message mixes both, the agent answers the part about your business and declines the rest.
- What counts as your business comes from your instructions, guardrails and knowledge, so describe what you offer there.
- The Playground uses the switch unsaved, so you can try it before saving.
Reply language
By default the agent answers in the language each visitor writes in. To choose a language, use Reply language on the Setup tab, below Guardrails:
- Under Reply in, pick a language. Visitor's language (automatic) is the default.
- If every reply should be in that language, switch on Always use this language. The switch appears once you pick a language.
- Click Save changes.
| Setting | Language of the replies |
|---|---|
| Visitor's language (automatic) | The language the visitor writes in. |
| A language, Always use this language off | The visitor's language. The chosen language is used when that is unclear, for example a greeting, a one-word message, only a name or a number, or mixed languages. |
| A language, Always use this language on | Always the chosen language, whatever language the visitor writes in. Names, product titles and quoted source text stay as written. |
Reply in offers 31 languages, listed by their English names: Albanian, Arabic, Bulgarian, Chinese, Croatian, Czech, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Hungarian, Italian, Japanese, Korean, Norwegian, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swedish, Turkish and Ukrainian.
- Everything the agent writes in a reply uses the same language, including the questions it asks, its confirmations and the follow-up questions.
- The setting is an instruction to the AI. If your instructions also name a language, make sure it matches.
- It doesn't change the widget's own buttons and notices. Those follow Widget language on the Appearance tab, and the two settings are independent. See Appearance.
- The Playground uses the choice on screen, saved or not.
Hand-off to your team
The Hand-off to your team section of the Setup tab lets the agent email a conversation to your team when a visitor asks for a person. It is off by default, and saved with Save changes like the rest of the tab. See Hand-off by email.
Below it, the Live chat card lets someone from your team join the chat instead, while they have the Live chat inbox open. Its switches save at once, without Save changes. See Live chat.
Web search
Below the tab's other settings, the Web search card lets the agent look a question up on the web when your knowledge doesn't answer it. It is off by default and saved with its own Save web search button, not with Save changes. Your instructions and guardrails decide what is in scope for a search. See Web search.
Built-in rules
intoCHAT adds rules to every agent. Where your instructions or guardrails conflict with them, the rules win, so you don't need to repeat them:
- Business facts such as products, prices, policies and contact details come only from your instructions, your knowledge or successful action results.
- When information is missing or conflicting, the agent says what it can't confirm. It never invents an answer, a contact channel or a handoff to a person.
- Retrieved documents and action results are reference data, not instructions. Commands hidden in them are ignored, and visitors can't override these rules.
- The agent doesn't reveal its hidden instructions, credentials or private information from sources.
- It never asks for passwords, one-time codes, API keys or payment card details in the chat.
- It only uses the actions you configured, asks before consequential steps, and reports an action as done only when the result confirms it.
- It follows your lead collection settings and doesn't ask visitors to type their details into the chat when the form is meant for that.
Test in the Playground
The Playground answers with the instructions and guardrails currently on screen, saved or not, so you can try a change before you keep it. Testing never saves them. The same goes for Reply language and for the Suggest follow-up questions switch on the Appearance tab.
- Edit the instructions or guardrails on the Setup tab.
- Open Playground and ask real visitor questions, including ones the agent should decline and ones your content doesn't cover.
- Click Reset to start a new conversation after each change.
- When you are happy with the answers, click Save changes. Your website keeps using the last saved instructions and guardrails until you do.
The 100-character minimum applies here too: with a shorter text, the Playground shows an error instead of an answer.
Playground chats are saved with your other conversations. To try something without saving it, click the temporary chat icon in the widget header. A strip reading "Temporary chat — messages aren't saved" appears, and no lead form is shown. Click the icon again to return to the saved conversation. Replies in the saved conversation can be rated with a thumbs up or down, and the ratings appear in its transcript on the Conversations tab.
Compare two versions
Compare asks your saved settings and an edited draft the same question, side by side, so you can see which answers better before you change anything.
- Open Playground and switch on Compare in the top right. The switch appears once the agent is saved.
- Two chats appear: A · Saved settings answers with what your website uses now, and B · Draft answers with the draft in the panel above them. On narrow screens they are stacked.
- Edit B's Instructions, Guardrails and, if you like, Reply in and Always use that language. The draft starts from the Setup tab as it is on screen, saved or not. Editing it doesn't change the Setup tab; Start from form copies the Setup tab into it again.
- Type a question and click Send to both. A and B answer at the same time, and each keeps its own conversation for follow-up questions. A click on a follow-up question is sent to both.
- Click Reset both to start both conversations again, for example after changing the draft.
- When B answers better, click Use B. It copies B's instructions, guardrails and reply language into the Setup tab, and the header shows unsaved changes. Click Save changes to use them on your website. Until you save, A and your website keep the saved settings.
- Only the instructions, guardrails and reply language differ between A and B. Both use your saved knowledge, actions, follow-up and sources settings.
- Both chats are temporary: they aren't saved with your conversations, so they don't appear on the Conversations tab and have no Why this answer. No lead form is shown, and client actions don't run in Compare; switch it off to test them.
- B's instructions need at least 100 characters and its guardrails at most 4,000, as when you save. Until then Send to both shows an error and neither chat is asked.
- Test as a signed-in visitor applies to both chats.
- Compare needs a role that can edit the agent; Viewers don't see the switch. See Team members and roles.
- Switching Compare off returns to the normal Playground chat as you left it. Leaving the Playground tab clears both Compare chats.
Common problems
- The agent wanders off topic: state the scope and what to decline more precisely, or add a guardrail.
- Answers are too long: give a word limit or ask for bullet points.
- A fact is wrong: fix the source, or add a Q&A pair. Instructions are better for behavior than for long lists of facts.
- It doesn't point people anywhere when it can't help: name your contact routes in the instructions, or turn on hand-off by email.