Communication — communication steps
The Communication category is the steps that talk to the lead . This is the main communication layer: sending prepared messages, AI replies, semantic routing, and delayed push messages for nurturing.
The Communication category is the steps that talk to the lead. This is the main communication layer: sending prepared messages, AI replies, semantic routing, and delayed push messages for nurturing. Most steps in any flow are from this category. The other families ( Changes — data changes, Terms — transition management, Tracking — sending events externally, Additional — auxiliary steps) serve the logic around this communication.
Communication is the bot’s “face” for the lead. This is where everything the lead sees in
Telegram lives: text, media, buttons. The conversion rate and the tone of the dialogue depend on the quality of these steps. Changes here are sensitive: change the wording of the greeting, and drop-off at the first touch changes. Add AI Reply instead of prepared text, and the texture of the conversation changes from robotic to “alive”.When to use which step
Send message — for prepared texts (greeting, instruction, link, fixed response). AI Reply — for smart replies to the lead’s message (dialogue, qualification, answers to product questions). AI Routings — for branching by the meaning of the reply (user intent, consent/objection, country, language). Push Messages — for nurturing silent leads in a standard funnel. Custom Push — a separate step with pushes on the waiting status (when you want controlled nurturing touches, not global ones).5 steps in the category
—Send Message — sending prepared text, media, files, and inline buttons. Does not require AI, does not cost tokens, and runs instantly. The basic step for any fixed message: greeting, instruction, affiliate link, disclaimer.
—AI Reply — a dynamic response from AI based on the lead’s message. Takes into account AI Settings — Bot Personality(the bot’s personality) + the dialogue history + the step settings (Goal, Requirements, Reply Addition). Uses tokens for each call.
—AI Routings — semantic routing after
AI Reply. Analyzes the lead’s response and selects a branch: “agreement”, “refusal”, “price question”, “irrelevant”. Does not send messages; only changes the route.—Push Messages — global flow-level pushes for nurturing silent leads. Run outside the funnel, but within the flow.
—Custom Push — a separate step with local pushes at a specific point in the funnel. Used when you need controlled nurturing touches at certain stages, not global ones.
Send Message vs AI Reply vs Push
Three key communication classes that are often confused. The main differences:
Property | Send Message | AI Reply | Push Messages |
|---|---|---|---|
What it sends | Prepared text | AI generates based on context | Prepared text with a delay |
Costs tokens | No | Yes (on every call) | No |
When it triggers | Immediately when the step is reached | After a message from the lead | After N hours of silence |
Typical use case | Greeting, link, disclaimer | Answering a question, qualification | Nurturing after 24/72 hours |
Use cases by segment
Communication steps are equally applicable across all segments, but the layout differs. A few real-world examples:
—Forex/crypto arbitrage: Send Message (greeting + affiliate link) →
AI Reply (qualification by interest) → AI Routings (agreement/refusal) → Send Message (registration instructions) → Push Messages (if silent for 24 hours).—E-commerce:
AI Reply (greeting, question about the need) → AI Routings (whether there is a specific product) → Send Message (product card + buttons) → Custom Push (if they did not place an order within 1 hour).—SaaS support:
AI Reply (diagnostics, answers based on the docs) → AI Routings (resolved / operator needed) → Operator step (if escalation is needed) or Send Message (if resolved).—Info business: Send Message (greeting + free material) →
AI Reply (interest qualification) → AI Routings (ready to buy / still thinking) → Send Message (payment by link) or Push Messages (delayed nurturing).FAQ
Section articles
Related
Was this article helpful?










