Skip to main content
A

AI Sales Agent module

Owners and sales managers

AI Sales Playbook

Train the AI Sales Agent with sales goals, recommendation rules, discovery questions, campaigns, escalation rules, guardrails, and owner-reviewed improvement suggestions.

Released

Overview

AI Sales Playbook is the control layer that teaches the AI Sales Agent how to sell. Business Knowledge tells the AI what the business knows; the playbook tells it what sales behavior the owner expects during customer conversations.

AI Sales Playbook turns owner-approved sales strategy into reusable rules for recommendations, discovery questions, objection handling, closing behavior, campaigns, escalation, and never-do boundaries.

The module is built for non-technical business owners. Instead of writing prompts, owners configure practical sales rules: which products matter, what to ask before recommending, how to handle hesitation, when to push a campaign, and when the AI should hand the conversation to staff.

Why It Matters

Many small businesses already know how they want to sell, but that knowledge lives in the owner's head or in informal staff training. The playbook gives Billions+ a structured way to capture that sales judgment and apply it across Telegram today and future chat channels later.

The result is an AI salesperson that can stay helpful without becoming pushy. It can ask for missing details, recommend from customer context, respect business guardrails, and keep the owner in control before any strategy reaches live conversations.

What Owners Configure

Sales Goal

Owners can set the main selling objective for the AI. Examples include converting more first-time customers, increasing average order value, promoting premium items, collecting booking details, or moving customers toward order completion.

Priority Products

Owners can mark up to ten product IDs as high-priority items. The AI uses these as selling preferences, not as permission to invent facts or override catalog availability.

Recommendation Rules

Recommendation rules define when the AI should suggest a product, category, tag, or priority product. A rule can require discovery first, so the AI asks for missing information instead of immediately showing products.

Qualification And Discovery

The playbook can define qualification fields such as budget, recipient, category, size, use case, location, or timeline. Discovery rules tell the AI which question to ask when the customer intent is relevant but the answer is still missing.

Objection Handling

Owners can prepare guidance for common objections such as price concern, delivery concern, quality concern, payment concern, stock concern, or customer uncertainty. The AI uses that guidance to respond naturally without sounding like a fixed script.

Closing Style

The closing strategy tells the AI how direct it should be when asking for the order. The current editor supports a configurable closing style and a maximum number of recommendations per turn.

Campaigns And Escalation

Campaign rules let owners give the AI time-boxed sales priorities, such as seasonal products, promotions, clearance pushes, or high-margin campaigns. Escalation rules define current-message triggers where the AI should guide the conversation toward human staff and, when configured, pause AI after the escalation reply is delivered.

Boundaries

Boundaries define what the AI must never do, such as offering unapproved discounts, promising unavailable stock, making unsupported delivery promises, or handling sensitive payment issues without staff.

How It Works

  1. The owner opens Sales Playbook from the dashboard.
  2. The page loads the selected business playbook draft and active published version.
  3. The owner edits the structured rules instead of writing a prompt.
  4. The owner previews the draft against a sample customer message.
  5. The owner saves the draft, publishes it, or disables the active playbook.
  6. Runtime conversations use only the enabled active published version.

Draft changes never affect live customers until the owner publishes. Disable saves the playbook as disabled and removes the active runtime version, so the AI falls back to its normal behavior without losing the saved draft.

Runtime Behavior

At reply time, the AI Sales Agent receives a compact Sales Playbook context. Product accuracy, business policy, order rules, and payment rules still take priority over selling behavior.

The relevance layer keeps the prompt focused. Recommendation rules can use the current customer message and the latest useful customer context for discovery follow-up. Objections, boundaries, campaigns, and escalation rules are evaluated from the current customer message so old conversation context does not keep triggering stale sales behavior.

Adaptive Suggestions

AI Sales Playbook now includes owner-reviewed improvement suggestions. The system can analyze recent sales events and show deterministic recommendations such as reviewing an objection, tightening an escalation rule, improving a boundary, or checking a frequently shown recommendation.

Suggestions do not edit rules automatically. Owners can accept, dismiss, or snooze them, then decide whether to update and publish the playbook. This keeps the AI improving over time while preserving business control.

What Comes Next

The next step is deeper product reference validation and richer sales intelligence. The direction stays the same: the AI should learn from successful conversations, suggest better playbook rules, and help owners improve sales without taking control away from them.

Related Pages

  • Feature: AI Sales Agent
  • Feature: Product Management
  • Feature: Order Management
  • How-to: Connect Telegram AI Sales Agent

Related content