If you are looking for a chatbot for WhatsApp, start with the business questions your customers already send you. WhatsApp is now live in Ultimo Bots: connect your business number, assign an AI agent, and let it answer incoming text messages using your business knowledge. You can review those conversations in the portal Inbox, pause AI, and reply yourself.
Ultimo Bots is an AI agent that answers customer questions from your business content and carries out configured skills, now through incoming WhatsApp text conversations as well as your website.
Your business AI now answers on WhatsApp
A customer asks whether you serve their area. Another wants to know what an appointment includes. A third has read your service page but needs someone to explain the next step. These are good starting points for a WhatsApp AI agent because your business already has the information needed to help.
The practical change is where that help happens. You can assign the agent you have prepared for your website to a connected WhatsApp business number. Its knowledge, behavior and configured skills remain the basis of its answers. You maintain the agent's business information, then put it where customers want to ask.
Choose a first purpose you can describe in one sentence, such as "answer questions about our cleaning services and collect requests for a quote discussion." That sentence gives everyone a way to judge the replies. A useful answer explains the service and offers the right next step. It does not have to settle every possible customer question.
For an existing customer-facing agent, review the content before connecting the number. Check your service area, current opening hours and contact details. The launch is a good reason to fix information that your staff have been correcting manually in messages.
Are WhatsApp bots allowed?
Yes. Meta's WhatsApp Business Messaging Policy allows automated responses within its customer service window and requires a clear route to a person, which can include phone or email. Businesses must keep their profile accurate, obtain the required opt-in for subsequent contact and honor opt-out requests. As of October 2026; practical guidance, not legal advice.
Ultimo Bots connects through Meta's official WhatsApp Cloud API. For your launch, give the number an obvious business identity and tell customers what they can ask. "Ask about our services and request a reply from our team" sets a useful expectation before someone starts typing.
Treat the phrase WhatsApp Business chatbot as a search category, then evaluate the actual service behind it. Your priority is an agent that knows your business, a connection you can test, and a person responsible for conversations that need judgment. Put those three things in place before promoting the number widely.
Set up your chatbot for WhatsApp
Prepare a completed agent first. Give it the business content it should answer from and configure any skills you want it to use. In the portal, open Deploy > Channels, choose WhatsApp, click Connect WhatsApp, and select the agent that should answer for the number.
The setup asks how you use WhatsApp today. Choose the option that matches your situation:
| Setup option | When to choose it |
|---|---|
| I want to connect a separate number | You have a number that is not already registered with WhatsApp |
| I use the WhatsApp Business app | You want to use the Business app connection path, subject to Meta's eligibility checks |
| I already use the WhatsApp Cloud API | You already manage an API number and have its Phone number ID and Access token; the WhatsApp Business Account ID is optional |
For the separate-number or Business app route, complete Meta's guided setup with access to the business account and chosen number. Follow the verification or approval steps Meta presents for that account. Existing Business app number continuity depends on eligibility in that setup, so make the choice based on your actual account rather than assuming any personal number will qualify.
Back in Ultimo Bots, check the number and assigned agent in Manage connections. Resolve any setup or activation warning you see. A connection confirmation is one checkpoint; an answer received by a customer is the meaningful test.
Send a fresh text from another WhatsApp account. Ask an ordinary question whose answer is in your knowledge base, check the reply on the phone, and find the conversation in Inbox. For a WhatsApp chatbot integration, this short test checks the whole customer experience instead of stopping at the connection screen.
Give the agent the answers customers need
Good source material starts with what customers need to decide. A service description should say who it is for, where it is available, what is included and how to take the next step. Give the agent those facts in its knowledge base, including the exact public links you want it to share.
| Customer question | Source to maintain | Useful answer |
|---|---|---|
| "Do you cover my neighborhood?" | Current service-area information | Explain the published coverage and ask for clarification if needed |
| "What is included?" | Approved service description | Describe the relevant inclusions in plain language |
| "How do I request a quote?" | Published request process and configured lead skill | Explain the process and collect the required details |
| "Can I speak with someone?" | Current public contact details | Give a usable contact route or take a follow-up request |
Write the exceptions beside the rule. If weekend visits need a separate agreement, put that next to your availability information. If a price applies only to a defined package, include the package conditions. This makes the source useful to both the agent and the person reviewing its answer.
Skills give the agent a configured way to act on a request. Where a skill needs an app action, connect the relevant app and include that action in the skill. The connection makes the app available; the skill defines how this agent uses it.
Start with the few steps you can verify yourself. You should be able to explain what information the customer provides, what the action does and what a successful result looks like. Test those steps after changing the skill or its source material.
A customer conversation from question to next step
Useful WhatsApp chatbot examples should show a complete customer request. Here is an illustrative conversation for the invented Cedar Lane Cleaning. Assume its approved content says it offers move-out cleaning, and its Lead Generation skill is configured to collect a name and email for the team to follow up.
Customer: Do you offer move-out cleaning?
AI: I'm Cedar Lane Cleaning's AI assistant. Yes, Cedar Lane offers move-out cleaning. Would you like the team to get back to you about it?
Customer: Yes, please. I'd like to discuss a quote.
AI: What is your name, and which email should the team use to reply?
Customer: Taylor Morgan, taylor.morgan@example.com.
AI: To confirm: Taylor Morgan, taylor.morgan@example.com, and you'd like to discuss a quote for move-out cleaning. Is that right?
Customer: Yes.
AI: Your request has been sent to the team.
The built-in Lead Generation skill asks for the required contact details, confirms them and forwards the request as a lead. Your team can use Leads to review the inquiry with its conversation. In this example, the customer supplied every detail that the agent read back.
Notice what makes the exchange useful: the service question gets an answer, the customer agrees to follow-up, and the team receives a clear reason to respond. The agent confirms that the request was sent. The quote discussion still belongs to the team, which can ask the questions needed to scope the job.
Before using this pattern, decide who will pick up new requests and what response expectation you can honestly publish. A good conversation ends with a next step your business can carry through.
What is the 24 hour rule on WhatsApp?
Under Meta's messaging policy, each incoming customer message opens or resets a 24-hour customer service window. Replies can use ordinary messages during that window; outside it, Meta requires approved message templates. This guide uses Ultimo Bots for incoming text conversations and replies during the open window.
Put that timing into your daily routine. When a customer has asked for help, review the conversation while it is current. If someone needs a later response, agree on a suitable contact route, such as the email address they provided for follow-up. A staff member replying manually also needs to work within the channel's messaging rules.
For example, a person who asks about a service on Tuesday afternoon may receive an AI answer and then leave a request for the team. Assign that request to someone rather than leaving it in the Inbox until the end of the week.
This is why your launch plan should include both good answers and clear ownership. The agent handles the first exchange; your team makes sure a request that needs attention reaches a person while the context is still useful.
Keep a clear route to a person
When you want to answer a WhatsApp conversation yourself, open it in Inbox, turn AI Auto-Response off, and send your text reply. Turn it back on when you want the AI to respond to new customer messages again. Practice that sequence with a test conversation before real customers depend on it.
Also put a working business phone number or support email in the agent's knowledge. A customer should be able to ask how to reach your team and receive a concrete answer. Choose contact hours and response expectations that match your staffing, then keep that information current.
Make the AI's identity clear in the messages themselves. In the AI Studio, choose Add behavior, then Enter manually, and add this instruction:
If none of your replies so far has said you are an AI assistant, say so briefly in this reply. The welcome message at the start does not count, since the visitor may not have seen it. If asked whether you are human, say plainly you are an AI.
Behaviors guide the agent's wording, so read the result in a fresh customer conversation. Check that the introduction is brief, the business name is clear and the human contact route works. Then try a direct question such as "Are you a person?" Good disclosure should remain clear when the customer asks again.
How much does a WhatsApp chatbot cost?
When comparing WhatsApp chatbot pricing, separate the AI service from Meta's messaging charges. Ultimo Bots counts WhatsApp AI responses against the agent's normal response allowance. One customer conversation can contain several AI replies, so estimate reply volume rather than assuming each chat uses one unit.
For plans purchased directly through the portal, these are the published USD prices per agent as of October 2026:
| Plan | Monthly billing | Per month, billed yearly | AI replies per month |
|---|---|---|---|
| Smart | $29 | $19 | 500 |
| Boost | $59 | $39 | 2,000 |
| Ultimo | $149 | $99 | 10,000 |
Meta's applicable messaging fees are separate and depend on its current market and message-category rules. Check the official WhatsApp Business Platform pricing for your business. An Ultimo Bots trial concerns the AI subscription; it is not a promise that every WhatsApp message has no platform charge.
Every paid plan starts with a seven-day free trial. Use the pricing page to compare the current allowances and decide which plan fits the agent's content and expected replies. During the trial, record how many replies your representative questions need, including the follow-up exchanges that collect and confirm contact details.
Include staff time in the decision too. Someone needs to review uncertain answers, update business information and respond to requests. Those are useful activities to measure during a pilot because they show what the channel will require when more customers use it.
Test Your Agent Free for 7 Days
Keep a first-week WhatsApp review log
Keep a short dated log alongside your normal launch notes. This is a team review practice you can maintain in a shared document. Record the question, what the customer received and the one change that would make the next answer more useful.
| Date | Customer question | Was the answer correct? | Source or behavior to update | Owner |
|---|---|---|---|---|
| First review | Question about service coverage | Compare with the current service area | Clarify a neighborhood boundary if needed | Content owner |
| Next review | Request for a person | Check the contact route and follow-up request | Correct contact details or response expectations | Support owner |
| End of week | Repeated question | Read several different phrasings | Improve the shared source, then test again | Agent owner |
Use actual conversations once they arrive. Until then, label your own test messages as tests. A repeated question is useful evidence: it may reveal that a service page leaves something unclear, or that the agent needs a more specific source. Fix the underlying information and ask the question again in different words.
At the end of the week, choose one improvement with a named owner. You might clarify a service condition, shorten an answer or make the contact route easier to find. Keep the initial purpose visible while you make those changes. An agent that answers your common questions accurately and gives customers a dependable next step is a strong foundation for the new channel.