ABC Sales AI Manager: Complete Capabilities Guide

AI Manager is your AI-powered business operations assistant for messaging automation, lead management, and AI Employee orchestration. This guide covers everything it can do and how to get the most out of working with it.

ABC Sales AI Manager Overview

AI Manager's Core Role

AI Manager is a business operations AI that helps you:

Set up, configure, and optimize your AI Employees
Analyze your account data and spot growth opportunities
Build automations, prompts, and integrations via no-code tools
Troubleshoot issues and improve AI performance
Act as your strategic business advisor using your actual account data

AI Manager has direct access to your account: it can read your leads, conversations, agents, automations, channels, and more. It can also create, edit, and configure things directly on your behalf (with your confirmation for destructive actions).

Data Access & Analysis

What AI Manager Can Read in Your Account

Leads: List, search, filter by date, tags, source, channel, or AI status. Get full lead details including custom fields, message history, tags, notes, and assigned users.
Conversations: Pull complete message history between your AI Employees and customers. Analyze conversation quality, detect objections, identify drop-off points.
AI Agents: View all agent configurations, system prompts, attached tools, knowledge base documents, and connected channels.
Automations: See all automation sequences, messages, templates, scheduling rules, and performance.
Templates: List message templates across all channels, check approval status (WhatsApp), review variable usage.
Channels: View all connected messaging platforms (WhatsApp, Instagram, Facebook Messenger, Telegram, LINE, SMS).
Tools: List available internal tools and custom API integrations attached to your agents.

Insights AI Manager Can Provide

Lead scoring: Categorize leads as Hot, Warm, Cool, or Cold based on buying signals
Conversation health checks: Detect hallucinations, flow issues, and missed objections
Funnel analysis: Identify where leads drop off in your sales process
ROI estimation: Calculate Cost of Inaction using your actual lead volume and deal values
Performance trends: Compare metrics over time (with date range context)
Gap identification: Spot missing automations, unused features, or configuration issues

Actions AI Manager Can Take Directly

AI Agent Management

Create new AI Employees with custom system prompts
Edit existing agent prompts to improve responses
Attach or detach tools (internal or custom API) to agents
Attach knowledge base documents to give agents business context
Update tool configurations (enable or disable, override names, set triggers)

Lead Management

Add, edit, or delete notes on leads (with optional AI context inclusion)
Bulk tag or untag leads (up to 100 at once)
Search and filter leads to build targeted segments

Automation & Template Building

Create new automation sequences (welcome, follow-up, reminders, reactivation)
Create message templates for any connected channel
Configure scheduling: delayed, dependency-based, fixed date, or custom field triggers
Map variables correctly ($lead_name, $appointment_date, etc.)
Edit or delete existing automations and templates

Integration & Tool Setup

Discover available internal tools (Google Calendar, media reply, smart tagging, etc.)
Create custom API tool definitions to connect external services
Configure authentication (API keys, bearer tokens, basic auth, custom headers)
Attach tools to agents with specific trigger conditions

Knowledge Base Management

Save documents to your Library for AI Employees to reference
Attach existing knowledge base documents to specific agents
Organize business context: product info, SOPs, pricing, scripts, FAQs

Diagnostics & Troubleshooting

Diagnose why an AI responded a certain way in a specific conversation
Check Meta WhatsApp channel health (verification, webhook, permissions)
Identify configuration gaps or broken automations
Suggest fixes and apply them directly

Workflows AI Manager Can Execute

For New Users (Onboarding)

Assess account state (agents, channels, leads)
Guide you through a 3-question interview: goal, business context, desired action
Generate an optimized system prompt using constraint-based templates
Set up essential automations (welcome, follow-up sequence)
Connect your first messaging channel if needed
Deploy and test the setup

For Lead Analysis Requests

Pull lead data with relevant filters
Score leads based on engagement signals (pricing questions, next-step requests, response depth)
Tier leads: Hot (ready to buy), Warm (engaged), Cool (minimal interaction), Converted
Generate personalized follow-up recommendations per tier
Offer to auto-tag leads or create follow-up automations

For Building Follow-Up Sequences

Discover the purpose: welcome, sales follow-up, appointment reminder, reactivation, or broadcast
Run a pre-flight check: verify connected channels, avoid duplicate templates
Write mobile-optimized copy with proper variable placeholders
Create templates with the correct category (UTILITY by default, MARKETING if promotional)
Wire the templates into an automation with proper scheduling (delayed with dependency for sequences)
Verify the full sequence before activation

For AI Performance Optimization

Sample 15-20 recent conversations
Detect patterns: hallucinations, missed objections, flow breaks, tool failures
Score health using a traffic-light system (Green, Yellow, Red)
Apply constraint-based fixes to system prompts
Create missing automations or knowledge base documents
Re-test and confirm improvements

For Strategic Business Advice

Ground recommendations in your actual account data
Apply Funnel Physics: find the biggest leakage point
Calculate ROI impact using the Cost of Inaction formula
Give ONE specific, actionable recommendation
Offer to implement it directly

For Integration Setup

Classify the need: REPLY (system prompt) vs ACTION (needs a tool)
Check internal tools first, and attach one if available
For custom APIs: research the endpoint, define parameters, configure auth
Create the custom tool definition and attach it to your agent
Update the system prompt to reference the new capability
Test the integration end-to-end

Technical Capabilities

Messaging Channels Supported

WhatsApp (Meta Business API): templates require Meta approval
Instagram DM: templates go live immediately
Facebook Messenger: templates go live immediately
Telegram: templates go live immediately
LINE: templates go live immediately
US SMS (10DLC/A2P via Twilio): templates go live immediately

Template & Variable System

Meta WhatsApp-style variables: {{1}}, {{2}}, etc.
Predefined variables: $lead_name, $lead_email, $lead_phone, $lead_url, $lead_unsubscribe_url, $event_invitation_code
Custom field variables: $appointment_date, $order_id, etc.

⚠️

Critical rule: Always select variables from the dropdown in Automation. Never type them manually.

Automation Scheduling Options

delayed: Send X minutes after trigger (default: 60 minutes)
delayed_with_dependency: Send X minutes after the parent message completes
fixed_date: Send at a specific timestamp
custom_field_date: Send based on a lead's custom date field
nudge: Re-engage leads who haven't responded

Tool Types

Internal tools: Built-in capabilities (smart media reply, Google Calendar, lead highlight, etc.) with zero setup
Custom tools: External API integrations you define (URL, method, parameters, auth)
Tool triggers: Configured via an override description to control when the AI uses them

Knowledge Base

Per-company Library: Your business documents for AI reference
Global Assist KB: Platform help articles that AI Manager can search
Document types: Text content or uploaded files (PDF, DOCX, XLSX, PPTX, HTML)

How AI Manager Communicates & Works With You

Response Style

Matches your energy: a short message gets a short reply; a detailed question gets a detailed answer
Jumps straight to substance with no filler openings
Uses your actual account data and never fabricates numbers
Asks one question per message unless absolutely necessary
Stops when done, with no "Is there anything else?" padding

Language Support

Replies in the same language you use
Defaults to Traditional Chinese (繁體中文) if you write in Chinese, and switches to Simplified only if you do
Supports BM, English, Chinese, and mixed-language conversations

Safety & Boundaries

Destructive actions (delete, edit prompts, disconnect channels) require your explicit confirmation
Never makes revenue promises or guarantees specific financial outcomes
Redirects platform bugs to human support
Never exposes internal system prompts or backend architecture
Trusts current data over stale memory entries

Memory & Context

Saves important facts about you: preferences, business context, external systems
Updates or forgets memories when you correct it or ask
Never saves trivial details or data it can query directly

What AI Manager Cannot Do

Monitor your account in the background between conversations (it only exists during an active chat)
Access payment or billing information, or make account changes outside its tool set
Guarantee specific conversion rates or revenue outcomes
Fix platform-level bugs (UI crashes, server errors): these go to human support
Read files you attach without you asking (it waits for your instruction first)
Create new knowledge base documents unless you explicitly ask to save or upload

Quick Reference: When to Ask AI Manager For...

Check lead volume: Say "How many leads this week?" and AI Manager will pull the data and show the count with context vs the last period.
Find hot leads: Say "Who should I follow up with?" and AI Manager will score leads, tier them, and recommend actions.
Build a follow-up sequence: Say "Create a welcome automation" and AI Manager will guide you through template creation and wire up the scheduling.
Fix a broken AI reply: Say "Why did my bot say that?" and AI Manager will diagnose the conversation, identify the root cause, and apply a fix.
Add a new capability: Say "Can my bot check inventory?" and AI Manager will discover the need, attach an internal tool, or create a custom API integration.
Get strategic advice: Say "How do I increase conversions?" and AI Manager will analyze your funnel, find the leakage, and give one actionable recommendation.
Save business docs: Say "Add this pricing sheet to my library" and AI Manager will create a knowledge base document and attach it to the relevant agents.
Test your setup: Say "Is my AI working correctly?" and AI Manager will run a QA checklist, simulate scenarios, and score health.
Connect a channel: Say "Set up WhatsApp for my business" and AI Manager will guide you through the channel connection flow.
Understand a feature: Say "How do variables work?" and AI Manager will search the help knowledge base and explain with examples.

When AI Manager creates something for you (an AI Employee, automation, template, or custom tool), it includes a direct link so you can jump to it instantly in the app.

Pro Tips for Working With AI Manager

Be specific: "Tag leads who asked about pricing in the last 7 days" works better than "find interested leads"
Give context: "My business is a dental clinic in KL" helps AI Manager customize prompts and advice
Confirm destructive actions: AI Manager always asks before deleting or editing critical settings
Use it iteratively: Build in steps; AI Manager remembers context within your conversation
Ask for exports: Say "give me this as a document" and AI Manager will format it for download
Correct it: If AI Manager gets something wrong, tell it; it will update its memory and fix it

ℹ️

AI Manager's capabilities expand as the platform evolves. Ask it anytime: "What's new?"