Free Voice-Activated Support Chatbot Template: Triage

The Voice-Activated Support template is a compact, twenty-node support assistant that takes an incoming question, sorts it into one of five support categories, asks one follow-up question in that category, and then lets the customer choose how they want the answer delivered: a written step-by-step guide, a video tutorial, an email, or a live human agent. It closes by asking whether the issue is actually resolved and loops back to the category menu if it is not. Every part of that path is built from Conferbot nodes you can open and rewrite, so the finished bot sounds like your support team rather than a generic assistant.

A note on the name before anything else, because it matters for expectations. The template is called Voice-Activated Support and its shipped greeting talks about a voice assistant, but the flow that ships with it is an ordinary text and tap conversation. In the template's own call settings, voice and video are both switched off, and there is no speech recognition node anywhere in the graph. What you get is a support-navigation flow: typed answers where a description is needed, tappable buttons everywhere else. Visitors on a phone can of course dictate into the text box using their own keyboard, and that works as it always does, but that is their device doing the listening, not the chatbot. If you were shopping specifically for speech recognition, this is not that template, and we would rather say so here than let you find out after you have rewritten all the copy.

What the template is genuinely good at is the thing most support teams actually need first: cutting the number of tickets that reach a human, without cutting off the people who need one. The five-category menu is broad enough to cover the bulk of inbound support at a software or subscription business, the single follow-up question in each branch captures the detail an agent would otherwise have to ask for, and the solution menu puts the choice of channel in the customer's hands. Someone who wants to fix it themselves gets the guide. Someone who is done trying gets the agent. Someone who is mid-commute gets it emailed.

It suits small support teams at SaaS products, subscription services, consumer electronics brands, and any business where the same handful of question types arrive over and over. It is deliberately lighter than its sibling technical support triage template, which drills much deeper into diagnostics, and lighter than the after-hours answering chatbot, which is built around capturing messages when nobody is at the desk. Pick this one when you want breadth and a fast path to either self-service or a human. Browse the rest of the library at support and FAQ templates.

The sections below walk the real conversation node by node, name the Conferbot node types that build it, and cover setup, customization, and channel fit.

The flow is a single block of twenty nodes joined by twenty-eight edges. It has one entry point and one exit, with a loop in the middle that lets a customer come back to the category menu as many times as they need. Here is the whole path in the order a visitor experiences it, using the wording the template ships with.

1. The greeting

A welcome-node opens with: "Welcome to our Voice-Activated Support Assistant. I'm here to help you with any questions or issues you may have. How can I assist you today?" This is sample copy. Rewrite it in your own voice, and if you are not enabling any kind of voice interaction, drop the phrase "Voice-Activated" so you are not promising something the flow does not do.

2. The open question

An ask-custom-question-node asks "Please state your question or the type of support you need." and stores the reply in the variable initial_query. This is a free-text box. Nothing in the flow branches on what the customer types here, which is intentional: the point is to capture the problem in the customer's own words, early, before any menu has nudged them toward a category. That raw sentence is the single most useful thing an agent can read at handover, and it is also the best raw material you have for deciding which categories your menu should actually contain.

3. The category menu

An n-choices-node then says "I understand you need help. Which category best describes your query?" and offers five buttons: Technical Support, Account Management, Billing and Payments, Product Information, and Other. The answer lands in query_category. This is the fork that shapes the rest of the conversation, and it is the first thing you should adapt: five buttons is the right number for most businesses, but they should be your five, named the way your customers name them.

4. One follow-up question per branch

Each of the five buttons leads to its own ask-custom-question-node, and each asks a differently worded question so the reply is usefully specific:

• Technical Support asks "Please describe the technical issue you're experiencing." into tech_issue_description.
• Account Management asks "What specific account-related assistance do you need?" into account_assistance.
• Billing and Payments asks "What billing or payment issue can I help you with?" into billing_issue.
• Product Information asks "Which product would you like information about?" into product_info.
• Other asks "Please provide more details about your query." into other_query_details.

All five branches then converge on the same node, so the graph stays flat and easy to edit. Note that the template asks exactly one question per branch. That is a design choice in favor of speed, and it is the main difference in depth between this template and a deeper diagnostic flow.

5. The acknowledgement beat

A plain message-node says "Thank you for providing that information. I'm processing your query to find the best solution." It carries no logic. Its job is pacing: it marks the end of the intake and the start of the answer, so the next menu does not land on top of the customer's typing.

6. The solution menu

A second n-choices-node asks "Based on your query, here are some potential solutions. Which would you like to explore?" with four buttons stored in solution_preference: View step-by-step guide, Watch a video tutorial, Connect with a human agent, and Receive solution via email. This menu is the heart of the template. Rather than guessing whether a customer wants to self-serve or talk to somebody, it asks them.

7a. Step-by-step guide

A message-node replies "Here's a step-by-step guide to address your issue: [Insert relevant steps here]". That bracketed placeholder is a to-do list for you. In practice most teams split this into one message per category branch so the steps match the problem, which is a small edit and a large improvement.

7b. Video tutorial

A user-redirect-node shows "Watch this video tutorial for a visual guide to your solution" with the button text "Click here to watch the video tutorial", pointing at a placeholder example.com URL. Swap in your real tutorial link, or one link per category.

7c. Human agent

A human-handover-node takes over with "Please wait while I connect you to a human agent who can assist you further." Customers waiting see "An agent will be with you shortly. Thank you for your patience." It is set to medium priority, automatic agent assignment, a five-minute maximum wait, and tags the conversation voice_support and escalation. If no agent picks it up in time, the fallback message offers a callback: "I apologize, all our agents are currently busy. Would you like me to schedule a callback?" Handover email and handover sound are both enabled, so your team is alerted rather than relying on someone watching the inbox. Live handover and the unified inbox are available on every Conferbot plan, including the free plan, so this branch works from day one. See live chat for how the agent side looks.

7d. Email the solution

An ask-email-node asks "Please provide your email address to receive the solution." into user_email, rejecting malformed addresses with "That doesn't appear to be a valid email address. Please try again." An email-node then sends to that address with the subject "Solution to Your Support Query" and a body that opens "Dear Valued Customer, Here is the solution to your recent query: [Insert solution details]". The chat transcript is attached automatically, which is the genuinely useful part: the customer gets a record of the whole conversation, placeholders or not. Replace the bracketed text before you publish.

8. Did that work?

All four solution branches rejoin at a yes-or-no-choice-node: "Has this solution resolved your issue?" with the labels "Yes, it's resolved" and "No, I need more help", saved to issue_resolved. This single field is the template's scoreboard. It tells you which categories deflect cleanly and which ones are wasting everybody's time.

9. Two closing messages

"Yes" leads to a message-node reading "I'm glad we could resolve your issue. Is there anything else I can help you with today?" "No" leads to "I'm sorry we couldn't fully resolve your issue. Let's try a different approach or connect you with a specialist."

10. The loop

Both closing messages arrive at a second yes-or-no-choice-node: "Would you like further assistance or to end this session?" with the labels "I need more help" and "End session", stored in session_continue. Choosing more help sends the customer back to the category menu, not back to the start, so they skip the greeting and the opening question and pick a new category immediately. Choosing to end reaches the final message-node: "Thank you for using our Voice-Activated Support Assistant. Have a great day!"

That loop is worth pointing out because it is the flow's only cycle and it is doing real work. A customer with two unrelated problems handles both in one conversation, and you get two clean category records instead of one muddled one.

The template is made from nine node types. Knowing which does what makes editing far quicker, because you can add or remove a step without unpicking the rest of the graph.

Two things are worth noticing about that inventory. First, there is no rating-choice or opinion-scale node, so the template measures success with a binary yes or no rather than a satisfaction score. If you want a score, add a rating-choice-node after the resolution check; the analytics dashboard will chart it, and the CSAT dashboard is available on Pro and above. Second, there is no ask-name node, so the bot never learns who it is talking to unless the customer chooses the email branch. If you want names on every conversation, an ask-name-node slotted between the welcome and the opening question costs you one extra step.

Every free-text and menu answer is stored in a named variable, which means the transcript your agent sees at handover already contains the customer's own description of the problem, the category they picked, and the delivery channel they wanted. That is the difference between an agent opening with "How can I help?" and opening with "I can see you have a billing issue with a duplicate charge."

The template fits businesses with a recognizable set of repeat questions and a small team answering them. Three patterns come up most often.

A small SaaS support desk

For a software product, the five categories map almost directly onto the queues a support team already runs: technical, account, billing, product questions, and everything else. The value here is the solution menu. A large share of technical questions have a written answer that already exists in your help center, and customers who are given a clear guide will often take it. The ones who will not are exactly the ones you want reaching an agent, and the handover branch routes them there in one tap with their description already captured.

A subscription or membership business

Here the billing and account branches carry most of the volume, and the email branch earns its keep: a customer asking how to change their plan or update a card wants the instructions somewhere they can find them again. Because the email node attaches the transcript, the customer keeps a record of what they asked and what they were told, which quietly reduces repeat contacts on the same issue.

A consumer hardware or product brand

Product Information and Technical Support dominate, and the video redirect branch is the one to invest in. Setup and troubleshooting are far easier to show than to describe, and pointing a customer at a two-minute video is usually faster than any written sequence. Wire one video link per category rather than using the single shipped placeholder.

It is a poorer fit in two situations. If your support questions need deep diagnostics, several conditional questions before you can even name the problem, look at technical support triage instead, which is built for that. And if your main problem is coverage rather than volume, nobody is at the desk in the evening, the after-hours answering chatbot is designed around message capture and callback scheduling in a way this template is not.

Nothing here requires code. The whole process is opening the template, rewriting the sample copy, wiring the two branches that point at placeholders, and publishing to a channel.

1. Fork the template

Open the Voice-Activated Support template from the support and FAQ library and create your own copy. You now have an editable flow of twenty nodes and can change anything in it. On the free plan you get one chatbot and six hundred conversations a month, which is enough to run this properly and see whether the categories are right before you pay anything. See pricing for the higher conversation and bot limits on Starter, Pro, and Business.

2. Fix the name and the greeting first

Before anything else, decide what you are calling this bot. If you are not doing voice, rewrite the welcome-node so it does not describe itself as a voice assistant. A greeting that over-promises is the fastest way to disappoint a customer in the first five seconds.

3. Rename the five categories

Edit the category n-choices-node so the buttons are your categories, in your customers' language. If you already run a ticketing queue, mirror those names exactly, it makes reporting comparable. Then edit the matching follow-up question in each branch so it asks for the detail that branch actually needs. Adding or removing a category means adding or removing one choice, one ask-custom-question-node, and two edges.

4. Fill in the three placeholders

The template ships with three things deliberately left blank. The step-by-step guide message contains "[Insert relevant steps here]". The video redirect points at an example.com URL. The email body contains "[Insert solution details]". All three must be replaced before you publish, and the step-by-step guide is the one most worth splitting per category.

5. Wire the handover

Open the human-handover-node and set the pieces that depend on your team: who is assigned, the maximum wait before the fallback fires, and the tags. The shipped tags are voice_support and escalation; change the first one if you have dropped the voice framing. Leave the handover email and sound alerts on so a waiting customer is not sitting in a queue nobody can see. Then decide what the callback fallback actually does, because as shipped it asks a question the flow does not act on. Adding an ask-phone-number-node and a calendar-node after the fallback turns it into a real booking.

6. Connect where the answers go

Wire the notifications and integrations you want before you publish, not after. Conferbot has sixteen native integrations including Gmail, Google Sheets, Google Calendar, Slack, HubSpot, Zoho CRM, Notion, Airtable, Stripe, and Webhooks, plus over two thousand more apps through Zapier. Google Sheets is the usual first choice, it gives you a row per conversation with the category, the customer's own description, the delivery preference, and the resolution answer. Slack or Gmail alerts suit handovers. Webhooks and the API need Starter or above. The integrations hub has the full list.

7. Pick a channel and publish

Choose where it runs, publish, and then walk the flow yourself as a customer. Take each of the five categories and each of the four solution branches at least once, which is a handful of minutes and catches every unreplaced placeholder.

8. Read the resolution field after a week

The first thing to look at is the split on "Has this solution resolved your issue?" broken down by category. A category with a poor resolution rate is telling you either that its guide is wrong or that it should be handing straight to an agent. That is a five-minute edit either way, and it is the loop that makes the bot better month over month. Chatbot analytics shows the drop-off point for every node.

A few changes repay the effort quickly.

Split the guide per category. The single shared step-by-step message is the template's biggest compromise. Give each of the five branches its own guide message and the self-service branch becomes genuinely useful instead of generic.

Send some categories straight to a human. Not every category should offer four options. A billing dispute or a suspected security problem rarely benefits from a written guide. You can route a category's follow-up question directly into the human-handover-node and skip the solution menu entirely for that branch.

Add a rating after the resolution check. A rating-choice-node gives you a satisfaction score alongside the binary resolved answer, and the two together tell you more than either alone: resolved-but-unhappy is a different problem from unresolved.

Ask for a name. An ask-name-node after the greeting lets every later message address the customer directly, and it puts a name on the record your agent picks up.

Use a knowledge base instead of hand-written guides. If your help content already exists, point the self-service branch at it. Conferbot's AI knowledge base can answer from your own documents, and you can connect OpenAI GPT, Anthropic Claude, Google Gemini, DeepSeek, or any of seventeen other providers to phrase the answer.

Make the callback fallback real. As covered above, an ask-phone-number-node plus a calendar booking node turns the busy-agents fallback from a polite dead end into a scheduled call.

Remove the Conferbot branding. Branding removal needs Starter or above; full white-label needs Pro or above.

Conferbot publishes to ten channels and none of them are locked behind a plan, so channel choice is about where your customers already are rather than what you are paying.

The website widget is the default and the best fit for this particular template, because three of its four solution branches, the written guide, the video redirect, and the email capture, work best on a device where the customer can immediately act on the answer. Put the widget on your help center, your pricing page, and your account settings pages rather than only the home page.

WhatsApp suits consumer and subscription businesses, where the conversation persists in a thread the customer already checks and a re-contact a week later lands in the same place. Instagram and Messenger make sense if your inbound support arrives through social DMs today. Telegram works well for technical and developer-heavy audiences.

For internal support desks, the IT help desk answering the same five categories for your own staff, Slack or Microsoft Teams is the natural home; the flow needs no changes beyond the copy. Discord fits community-led products, and there is also LINE and a mobile SDK. Running the same flow on several channels at once is normal, and omnichannel keeps every conversation in one inbox so an agent picking up a handover does not care which channel it came from.

If you are building a wider support stack, the survey templates pair well with this one for post-resolution follow-up, and lead generation templates cover the presales questions that inevitably arrive in a support widget. In 2026 the practical advice is still the least glamorous: pick one channel, get the five categories right, then expand.