WhatsApp AI Sales Assistant: From Inquiry to Follow-Up
See how a WhatsApp AI sales assistant connects product answers, lead qualification, customer memory and follow-ups, with human control over quotes and handoff.
A sale on WhatsApp rarely ends with the first answer. The customer asks about a product, discusses quantity and delivery, compares a quote, then returns with another question. WhatsApp AI Pro connects those steps: AI answers from your business information, builds customer context, tracks commitments and prepares follow-ups, while your team handles the decisions that need a person.
This guide follows one sales conversation through that process. It is the practical companion to one AI brain for the customer lifecycle; the AI CRM tutorial shows the daily controls in detail.
Follow one buyer from inquiry to the next step
Illustrative example: Cedar Supplies sells replacement filters. A fictional buyer, Maria, needs 200 packs and wants to understand delivery before deciding. The dialogue below demonstrates a workflow; it is not a customer case study, a benchmark or a guarantee of an identical AI response.
| Stage | Customer or team input | AI work | Your decision |
|---|---|---|---|
| Product inquiry | “Do you supply replacement filters in packs of 20?” | Prepare an answer using saved product information | Check that the product and pack size match |
| Qualification | “I need 200 packs for delivery next month” | Capture quantity, timing and related intent signals | Review the customer record and any inferred stage |
| Quotation | The buyer asks for a delivered price | Use provided pricing and ask for missing delivery facts | Confirm freight and any customer-specific terms |
| Commitment | You say “I'll confirm freight on Friday” | Detect the promise and surface it when due | Obtain the cost, then approve the follow-through |
| Follow-up | The buyer has not replied to the quote | Prepare a message based on the unresolved question | Approve, defer, or allow eligible routine automation |
| Human handoff | The buyer requests an exception or discount | Escalate according to your configured rules | Negotiate and confirm the commercial decision |
Many Conversations at Once, Without a Bigger Team
AI can assist across customer conversations while keeping each customer's context available for review. Use Solo Suggest for reviewed drafts, Solo Auto for permitted routine replies, or Watch when you want to write messages yourself while CRM and translation assist in the background. In Team mode, AI drafts go to human reps.
For Maria's inquiry, start in Suggest. You can see the original question, the proposed reply and the customer context before approving. The Solo vs Team guide explains the controls; reply speed and capacity depend on the computer, connection, AI service and messaging conditions.
Qualifying by Interest, Not by Gut Feel
As the conversation develops, the AI extracts information such as the customer's company, needs, budget signals and deal stage. It can assign an A–D intent tier and a status such as Active, Hesitating or Cooling. These are assessments for your review, not proof a customer will buy.
In the example, Maria's quantity and delivery deadline are useful facts. A price question alone does not establish her budget or authority to purchase. Read the underlying conversation and correct the record when the interpretation is wrong.
That record supports the daily “who needs attention” briefing. Your rep can return to the delivery question without rereading every chat. See customer records and the daily plan.
Quotes That Come From Your Price List, Not the Model's Imagination
Provide product names, pack sizes, prices, quantity tiers, minimum orders and delivery terms through company knowledge setup. The AI uses those facts to prepare pricing answers. Check units and conditions as carefully as the final number: a per-pack price and a per-unit price are not interchangeable.
For Maria, an appropriate reviewed draft might be:
“We can quote the filters using our current price list. Which city and postcode should we use to confirm freight for the 200 packs?”
If freight is missing, the AI should ask for clarification or route the question instead of filling in a plausible charge. A focused clarification can ask you for a missing business fact; inspect the saved answer before relying on it later. Negotiated discounts and exceptions belong with your sales team.
Keep the promise attached to the quote
When you write “I'll confirm freight on Friday,” the AI can detect the commitment and add it to the promises ledger. A due promise can surface in your daily briefing and generate a follow-through draft. Promise follow-ups always require approval, including when routine replies use Auto mode.
You must still obtain the freight price. The assistant tracks the commitment and prepares communication; it does not carry out every external task mentioned in a chat.
Product Questions, Stock Checks, and Simple Orders
A catalog tells the AI what you sell. For stock or order status, the AI can look up your own spreadsheet; replies quoting a price or delivery time from it wait for your approval.
When Maria is ready, the AI can help collect the product, quantity, destination and terms into an order summary. A person confirms the order, takes payment and enters it into the order or accounting system. This workflow does not create payment links or write orders into an external system.
Follow up on the conversation, not just the calendar
A useful follow-up addresses what prevented the next step. If Maria was waiting for freight, asking whether she saw the confirmed freight amount is more useful than sending a generic reminder.
An illustrative draft, after you have actually confirmed and sent the cost:
“Hi Maria, did the freight estimate answer your delivery question for the 200 packs of filters? Let me know if there is another detail you need before deciding.”
Solo Suggest keeps routine follow-ups as drafts. Solo Auto can send eligible routine follow-ups during the customer's local business hours. You can defer a customer you are handling personally. Review CRM follow-up controls before choosing a mode.
Reaching out to old customers about a new catalog or restocked item is a separate win-back campaign that you start. Its messages are generated at send time and cannot be previewed individually. It is not an inventory-triggered back-in-stock notification or the same approval workflow as a CRM draft.
The Handoff: Rules, Then a Clean Claim
Define when a human should take over: the customer asks for a person, requests a discount, raises a complaint, or needs an answer outside the available business facts. Test the rules with realistic conversations before delegating routine traffic.
In Team mode, human reps handle AI drafts through Telegram, Slack, WeCom or the Enterprise Inbox. The customer stays in the same WhatsApp thread, while the rep can review the history and continue the conversation. See the Inbox tutorial for claiming and reviewing messages.
Escalation Triggers Worth Writing Down
- A direct request for a human.
- Negotiated discounts, custom payment terms or commitments above your chosen threshold.
- A complaint, refund request or dispute.
- Conflicting or missing product, price or delivery information.
- Any action that requires a person to confirm something outside the conversation.
For Maria, a request for a special credit arrangement is a clear handoff point. The AI can keep the context organized while your team makes the decision.
Improve the next conversation from this one
If the draft missed a unit or sounded too pushy, correct it and explain why. Fix an incorrect price in company information; use a lasting reply preference for a recurring writing requirement. Feedback can also produce learning improvements that apply automatically when the evidence supports them, and each one can be revoked.
This connects the sales workflow back to the AI's next reply. Improvement should be visible in the work you review, rather than assumed because the assistant has processed more messages.
What This Setup Deliberately Doesn't Do
- It does not autonomously negotiate exceptions, collect payment or write an order into your accounting system.
- It does not book calendar slots; a person confirms an appointment.
- It does not turn a lead score into a verified purchase intention.
- Desktop automation needs a running, connected computer. Choose review and takeover settings appropriate to your business.
Frequently Asked Questions
How does the AI decide a lead is qualified?
It assesses signals from the conversation, such as quantity, timing, budget discussion and buying concerns, and records an intent tier and stage. Your team should review that assessment and correct it. It is a prioritization aid, not a prediction of a guaranteed sale.
Can it send personalized quotes automatically?
Solo Auto mode can send routine pricing replies using supplied business facts. Use Suggest when a price needs approval, and keep negotiated terms with a person.
When exactly does a human get pulled in?
Configure escalation requirements for complaints, exceptions and questions outside the available facts. Team mode routes drafts through reps in Telegram, Slack, WeCom or the Enterprise Inbox. Test those controls on the scenarios your team actually encounters.
Does the customer notice the handoff between AI and human?
The conversation continues on the same WhatsApp number with its history available. The rep should read that history so the customer does not need to repeat the question.
Can it book appointments or demos by itself?
The workflow described here qualifies and routes the inquiry; a person schedules the appointment.
How many conversations can it handle at once?
The assistant supports work across conversations, but there is no universal throughput guarantee. Available compute, network conditions, AI service response time and WhatsApp messaging conditions affect capacity. Test your expected workload in Suggest mode.
Can it check stock and take an order?
It can answer stock questions as described above and help prepare an order summary. A person confirms the order, collects payment and records it in the relevant business system.
Will it negotiate a discount on its own?
Configure negotiation and exceptions as handoff cases. A published volume tier can be supplied as business knowledge; inventing a special discount is a commercial decision for your team.
Start with a reviewed product reply, then inspect the customer record and next action. That is the useful test of an AI sales assistant: whether it helps you keep the whole relationship moving.
Related guides: AI CRM · Full-workflow AI · WhatsApp AI agent comparison.