Typical MAX bot scenarios: lead generation, support and booking (Booking)

Companies often come to integrators with the request: “Make us a bot so that it can do everything.” This is a recipe for missed deadlines and disappointment. Practice shows that 90% of the economic effect (ROI) from intelligent agents (based on MAX or other LLMs) comes from just three highly specialized scenarios.

Let’s look at where to start with automation, how it works “under the hood” and what KPIs to measure success by.

---

1. Scenario: Deep lead generation (Pre-sales Qualification)

Button bots (“Press 1 for price, 2 for delivery”) are dead. Smart agents engage in natural dialogue and act as tireless primary sales managers.

Process architecture:

1. Grip: The client switches from advertising (UTM tags are embedded in the URL, which the bot invisibly parses).
2. Qualification: The bot finds out pain points, budget and deadlines (BANT framework), communicating in free text.
3. Enrichment: The bot searches for the client’s company by tax identification number through external APIs (DaData / Contour).
4. CRM routing: The collected data is formed into a blank Lead card (amoCRM / Bitrix24), and the task is assigned to a specialized Senior manager.

KPI for business:

• Conversion from “started a conversation” to “qualified lead in CRM”.
• The amount of hours saved by the sales department (excluding work with “junk” leads).
• Cost of lead acquisition (CPA) within the channel.

---

2. Scenario: Customer support (Support L1 / L2)

The fastest way to recoup the development of a bot for e-commerce or SaaS is to give it the first line of support (L1) routine. Unloading the contact center even by 30% provides enormous financial benefits.

Process architecture:

1. RAG analysis (Retrieval-Augmented Generation): The user asks a question. The bot searches for an answer on the fly in the internal knowledge base (Confluence / Notion) and synthesizes a clear answer, rather than just sending the client a link to a 10-page PDF.
2. API requests in ERP: The bot is integrated via REST API with 1C or internal ERP. To the request “Where is my order #664?” he makes a request to the database, pulls out the status, the logistician’s track number and the delivery date.
3. Escalation (Handover): The key part. The bot analyzes sentiment (Sentiment Analysis). If the client is angry, the robot becomes silent and instantly transfers the dialogue to the Senior operator marked (ALARM).

KPI for business:

• Deflection Rate is the percentage of requests completely closed by a bot without human help.
• Time to Resolution (TTR) is the average time to resolve a user’s problem.
• CSAT (Customer Satisfaction Score) - assessment after completion of the dialogue.

---

3. Scenario: Booking / Appointments

The scenario is ideal for medicine, car services, service industries and B2B business meetings.

Process architecture:

1. Slot synchronization: The bot parses available slots from Yclients, DIKIDI or Google Calendars of doctors/specialists (taking into account vacation schedules and overlaps).
2. Booking via WebApp: For complex choices (choosing a master, chair, service), the bot sends a Mini App (Web App), where the user selects parameters in a convenient visual interface without leaving the messenger.
3. Holding and reminders: A preliminary reservation is being created. The bot sets up a CRON task to send a push reminder 24 and 2 hours before the visit, asking for confirmation (“Are you coming?”). In case of failure, the slot is automatically defrosted in the system.

KPI for business:

• Reducing the No-Show rate (failure to show up without warning).
• Increase in the total number of bookings outside of call center business hours (24/7 effect).

---

How to choose the first scenario to implement?

Don't implement everything at once. Use the bottleneck rule. Start with a process that meets one of the criteria:

• You are losing real money there right now (leads are cooling due to the long response time).
• The largest number of people working there do Ctrl+C / Ctrl+V operations.
• You can reliably measure the “Before” and “After” process metrics (over a period of 3-4 weeks).

Specialists NBM-IT develop and implement intelligent agents, integrating them directly into the business logic of your ERP and CRM systems. Get a pre-project analysis and calculation of the cost of automation for your department.