Bot modules and blocks
Before you build a qualification bot for a client, it helps to know what you're actually building.
Last updated on
Before you build a qualification bot for a client, it helps to know what you're actually building. In Chatfuel, the bot isn't a flowchart you draw on a canvas — it's an AI agent assembled from five modules: what it knows, what it asks, where and when it answers, when it hands over to a human, and which channel it lives on. Get those five right and the funnel works; when something breaks, it's always one of the five.
In the product, those modules live in two places. What the bot knows is the Knowledge Base. Where and how it behaves is Automations — at Fuely AI → Automations, one automation per traffic source, each with its own settings. You can configure every module either right there in the dashboard or by describing what you want to the AI Co-Worker in chat — both edit the same setup, so use whichever is faster for the change at hand.
The five modules
| Module | What it controls | Where it lives |
|---|---|---|
| Knowledge Base | Everything the bot knows: the client's company profile, catalog of services or products, FAQ | The Knowledge Base |
| Qualification fields | The handful of things the bot finds out from every new lead — budget, timeline, service wanted — and how it asks | Collect contact info + Additional instructions on the automation |
| Reply settings | Where the bot answers (which sources are on) and when (always, or only outside working hours) | The per-source automations and their When AI replies setting |
| Human handoff | The rule that moves a chat to the client's team once the lead is qualified — plus a push notification in the Chatfuel mobile app | Switch to human on the automation |
| Channel | Where the bot talks: the client's WhatsApp, Instagram, Facebook, TikTok, or website widget | Channel connections in Settings |
Knowledge Base — what the bot knows
The single biggest quality lever. It holds the client's business profile (name, hours, address, contacts, payment options), a catalog of what they sell, and an FAQ. A thin Knowledge Base produces a vague bot; a well-fed one answers like the client's best salesperson. This is where most of your setup time should go — here's exactly what to feed it.
How the bot talks — its role, tone, house rules — is behavior, not knowledge: that goes into Additional instructions on the automation, on Main bot for everywhere or per source where a client wants a different voice for a different audience.
Qualification fields — what the bot asks
Usually three fields, chosen for the client's niche: a real estate bot asks district, budget, and timeline; a clinic bot asks service, main concern, and preferred date. The fields themselves live in the automation's Collect contact info setting; the conversational style — one question per message, not a form — is wired into its Additional instructions. The AI weaves the questions into a natural conversation and tracks the answers on the lead's profile. How to pick the right fields.
Reply settings — where and when
Two decisions, both made per automation:
- Where. Every traffic source has its own automation, and enabling it is what turns the bot on there. "Answer leads from ads but not direct messages" isn't a special mode anymore — it's just the ads source on and the DM source off. On comment and ad sources you can go further with custom automations: named rules for specific keywords, posts, or ads — a comment funnel for the client's launch post, a different one for the discount reel. To target a rule at a particular ad, paste its ID from Meta Ads Manager — or just paste the full Ads Manager URL and Chatfuel extracts the ID itself.
- When. Each automation's When AI replies setting: Always, or Outside working hours. For a lead-gen funnel the usual answer is "always" — but there are cases for the other mode.
Human handoff — where the bot stops
The bot's job ends when the lead is qualified. The Switch to human setting then moves the chat to the client's team inbox, with the collected answers summarized on the lead — and the Chatfuel mobile app pings the phone the moment it happens. That's the "lead" at the end of your ads → bot → qualification → lead pipeline: a warm, pre-qualified conversation a human picks up mid-flow.
Channel — where the bot lives
Each connected channel in the client's company — WhatsApp, Instagram, Facebook, TikTok, the website widget — feeds its conversations to the same AI agent. Connect a second channel and the same Knowledge Base and the same Main bot defaults apply there too: you configure the brain once, then override individual settings per source only where the client needs different behavior.
Note: one sentence still sets up all five. Tell the Co-Worker "Set up WhatsApp" and it walks you through every module in one guided session: it learns the business, proposes qualification fields tailored to the client's niche, turns the AI on, arms the handoff, and lets you test the bot in the Test chat before anything reaches a real lead. The channel-connection step at the end is the only part you finish by hand.