Chatbot Demo: Try Real Customer Questions

To try a chatbot demo, open the chat bubble on this page and ask about Ultimo Bots, such as which plans include live chat. Use the questions below to examine the answer and its next step. For questions about your own business, prepare your own agent with your content; Ultimo Bots offers a free seven-day trial of its paid plans. A useful demo lets you inspect what happened, from the first answer to the request your team receives.

Test your own customer questions

Ultimo Bots is an AI agent that answers visitors using your business knowledge, captures follow-up requests through configured skills and supports human takeover on your website and chat link while a teammate is online.

Try the chatbot demo on this page

Start with something you can verify. Ask the company agent, “Which plans include live chat?” Then try, “What information would you need to answer questions about my business?” Read the response, follow any relevant product link and compare it with the information on the linked page.

This public conversation is about Ultimo Bots. Your opening hours, service conditions and customer requests belong in a separate evaluation using your own agent. Bring one question customers frequently ask and the approved answer your staff would give. Those two pieces make the next demo much more informative.

Keep the difference between demonstration formats clear:

Use the public demo to get a feel for the conversation. Use the owned test to judge fit for your business. An elegant answer about someone else's service is a reason to investigate further; your actual customer questions provide the stronger evidence.

How does a chatbot work?

For a business demonstration, separate three things: the facts available to the agent, the instructions guiding its response and the actions it can perform. A reply may combine all three. You need to know which part produced the result so that a future correction has a clear home.

In Ultimo Bots, the knowledge base supplies business information. Behaviors guide how the agent responds. Skills define a task, such as collecting a request for someone to follow up. If that task needs an app action, the app must be connected and its action included in the skill.

For an AI agent demo, ask the person presenting it to show the source behind a factual answer and the configured skill behind an action. You are checking whether the system can support a customer need you recognize. You are also checking whether your team can understand and maintain the setup.

Notice the introduction too. Ultimo Bots shows an AI notice in the website widget and chat link. Read the opening as a visitor: the business identity, the AI's role and the help offered should be clear. Save any confusing wording for the person responsible for the agent's instructions.

Can you give me an example of a chatbot?

A useful example is a visitor asking a bicycle repair shop about a service and then requesting an estimate discussion. The conversation below is illustrative, not a recorded demo or a customer case study. The invented Northline Bicycle Repair has supplied these facts: its standard tune-up includes brake and gear adjustments; replacement parts cost extra; an estimate requires a staff assessment.

Its Lead Generation skill is active and requires a name and email. The visitor supplies both before the AI confirms them.

Visitor: Does the standard tune-up include new brake pads?

AI: The tune-up includes brake adjustments. Replacement parts, including brake pads, cost extra.

Visitor: I'd like someone to discuss an estimate with me.

AI: What is your name, and which email should the team use?

Visitor: Alex Rivera, alex.rivera@example.com.

AI: To confirm: Alex Rivera, alex.rivera@example.com, and you'd like to discuss an estimate for your bicycle. Is that right?

Visitor: Yes.

AI: Your request has been sent to the team.

For a customer service chatbot demo, inspect the distinction preserved in the first reply. Brake adjustment is included; new parts are a separate cost. An answer that simply says “yes, brakes are included” could sound helpful while giving the visitor the wrong expectation.

The final line has a similarly specific meaning. The request has been sent. A staff member still assesses the repair and discusses the estimate. Your own demonstration should make the completed action and the team's next responsibility just as easy to distinguish.

Ask the same question in two different ways

Try “Are replacement pads included?” and “Will I pay extra if you fit new pads?” Both ask about the same condition. Compare the business facts in the answers, rather than expecting identical wording. The cost distinction should survive the change in phrasing.

Then ask a follow-up in the same conversation: “What about the adjustment?” You are checking whether the reply connects that short question to the tune-up you were discussing. Repeat the original question in a fresh conversation as well, so the earlier explanation is not doing all the work.

Use language customers actually use. Someone may write “my brakes squeak” before asking about a service. A useful answer should help them understand the published service without inventing a diagnosis or a price. Keep your own judgment tied to the exact information you supplied.

Open the conversation on a phone and enlarge the text in your browser. Try reaching its controls with the keyboard. W3C's accessibility Easy Checks covers initial checks including keyboard focus and zoom, while explaining that these checks are not a complete accessibility evaluation. Record what you observe in the actual website experience you plan to offer.

Try the question your website does not answer

For Northline, ask, “Can you guarantee the bike will be ready tomorrow?” The supplied facts contain no turnaround promise. This is a useful test because a confident invented deadline could affect a customer's plans. You want the agent to recognize the information gap and offer a suitable next step.

Ultimo Bots instructs the agent to look up company facts, ground its answer in returned information and use real links from its tools. Read the result yourself. Those instructions are a reason to test carefully, and the observed answer is the evidence you should keep.

Try a gentle push: “I thought you promised next-day repairs.” The desired response should preserve the uncertainty instead of accepting an unsupported statement as company policy. Record any invented deadline verbatim so the issue can be reproduced and reviewed.

If a real policy exists elsewhere, add the approved source and try again. If staff judgment is required, keep that distinction visible and test the follow-up request. This separates missing content from a question the business should deliberately route to a person.

Follow one request beyond the chat window

Run a clearly labeled test request in your own agent with contact details you control. Tell the colleague receiving it that it is a test. The built-in Lead Generation skill collects a name and email by default, confirms the details and forwards the request. Additional fields depend on how you configure the skill.

Now open Leads in the portal and inspect the saved request and its conversation. Does it contain the contact information the visitor actually supplied? Can your colleague understand why the person wants help? The visible acknowledgment and the saved result should tell the same story.

Repeat with a correction before the request is sent. For example, provide one test email, then explicitly replace it with another address you control. Check the confirmation and saved result for the corrected detail. Also try declining to provide an email and observe how the conversation continues.

When evaluating a connected action, inspect its destination too. An app connection makes the app available; a skill must explicitly use its action. Ask what will be written or sent, and choose an authorized test that your team can identify afterward. Use the AI Studio to review the intended skill before running it.

Check your first captured request

Ask for a person, with someone online

Arrange this part of your own demo with a colleague. On the website and chat link, Ultimo Bots supports live chat takeover while a teammate is online. Have the colleague switch online in the portal, then ask to speak with a person from the visitor's chat.

Wait for the colleague to join and exchange a message in the same conversation. Check that the colleague can read the preceding discussion and that the visitor can tell a person has joined. While the teammate is participating, the AI stops replying. Your observation should include the actual joined conversation, not just an acknowledgment of the request.

Test the other staffing situations separately. With nobody online, the AI remains available and can collect a follow-up request through the lead skill. If someone is online but nobody joins a requested takeover within your configured wait time, the configured offline message appears. These are different experiences, so give each its own note.

The public company demo may have different staffing from your test. Use your own setup for this check, and ask the colleague whether the conversation gave them enough context to help. That answer tells you something useful about everyday service, beyond whether the chat window changed state.

Is chatbot free?

Demo access and ongoing service are separate buying questions. Every new Ultimo Bots agent starts on a small one-time free allowance. Paid plans start with a free seven-day trial, and checkout collects a payment card. Check the selected plan and billing terms before starting that trial.

As of October 2026, direct portal plans cost $29 for Smart, $59 for Boost and $149 for Ultimo per agent with monthly billing. They include 500, 2,000 and 10,000 AI replies per month respectively. The pricing page also lists yearly billing and each plan's knowledge, skill and behavior allowances.

During your evaluation, count the AI replies in a representative conversation. The Northline example needs several replies to answer the question, collect details, confirm them and acknowledge the request. One customer inquiry can therefore use more than one response from the allowance.

Include your team's work in the evaluation. How long does it take to check an answer, correct a source and respond to a captured request? Use what you observe to choose a manageable first use. A trial is more useful when it ends with a service your team understands and a realistic view of its upkeep.

Keep the evidence from your demo

Save a small collection of annotated conversations outside the product, alongside your evaluation notes. Each should contain the visitor's exact words, the observed answer, the approved source and any result you checked in the portal. Include a date so later changes have a clear comparison point.

Choose examples that reveal different things. One should show a service condition preserved through two phrasings. Another should show how missing information was handled. A third should include a completed request and the record your colleague inspected. Add the joined human conversation if takeover is part of your planned service.

Beside a weak result, write the smallest concrete correction: clarify the parts policy, add the approved turnaround wording, or review how the skill collects an email. After the change, keep both versions of the conversation. That pair makes the improvement visible without relying on someone's memory of a good demonstration.

Finish by bringing one unresolved customer question into your own trial. It might be the exception that staff explain every week or the request your current form handles poorly. Use that question, its real source and a colleague's judgment to decide what your first agent should do.

Try your agent free for 7 days