Customer Service Chatbot

A customer service chatbot that answers from your content, not a script.

SupportHQ is a customer support chatbot trained on your docs, your website, and your past answers. It replies on your site, in Telegram, and in Discord. It says so when it does not know, and it passes the conversation to your team with a summary.

Start free trial Book a call

Answers only from content you approved
Hands off to a person with a summary
Website, Telegram, and Discord
Flat monthly price, no per-seat fees

Three checks before you buy any support chatbot

Most AI chatbots for customer service look the same in a demo. These three questions separate them.

Grounding: where do answers come from?

Ask what the bot does when your docs do not cover a question. The right answer: it says so.

Handoff: what happens when it is stuck?

Ask who gets an alert, how it arrives, and what context they see when they open the conversation.

Pricing: what do you pay for?

Ask whether the price grows with seats, with resolutions, or with conversations.

Rule-based vs AI

Two kinds of customer service chatbot

A rule-based chatbot follows a decision tree. Someone draws the flows, writes each reply, and maps keywords or buttons to branches. It is predictable. It also breaks the moment a customer phrases a question in a way the tree did not expect.

An AI chatbot trained on your content reads the question, searches your knowledge base, and writes an answer from what it finds. Nobody draws flows. The risk moves elsewhere: an AI bot that is not grounded can sound confident and be wrong. That is why grounding is the first thing to check.

Aspect	Rule-based chatbot	AI chatbot trained on your content (SupportHQ)
How it answers	Picks a pre-written reply from a branch of the tree.	Writes a reply from the sources you approved.
Setup	Someone draws the flows and writes every reply by hand.	Upload docs, sync your website, or write articles.
A question nobody planned for	Falls back to "I didn't understand" or a menu of buttons.	Searches your content. If the answer is not there, it says so and escalates.
When a policy changes	Edit every flow that mentions it.	Update the source document. The next answer uses it.
Other languages	Usually one tree per language.	Replies in the language the question arrives in.
Where it fits best	Short, fixed tasks with few paths.	Open questions about your product, policies, and setup.

A rule-based bot still makes sense for a narrow, fixed task. For open questions about a product, the second kind usually fits better, as long as it stays grounded and hands off well. SupportHQ works as an AI agent: it decides when to answer, when to stay quiet, and when to escalate. Read what an AI support agent is, or AI agent vs chatbot for a deeper look at the difference.

How it works

From your docs to a working chatbot

Most teams are live the same day.

1) Add your content

Upload files, sync pages from your website, write articles, or turn answered conversations into knowledge.

2) Put it where customers ask

Paste one script tag to add the chat widget to your site. Connect Telegram or Discord from the dashboard.

3) Review and improve

Read every conversation in one inbox. Fill the gaps the agent reports, and step in whenever you want.

In depth

The buyer's checklist, and how SupportHQ answers it

Use these points to judge any customer service chatbot, including ours.

What a customer service chatbot is expected to do today

Many customers now expect more than a bot that only shows a menu. They want a direct answer to a typed question, at any hour, in their own language. They expect the answer to match your actual policy. And when the bot cannot help, they expect a person to pick up without asking them to repeat everything.

Your team has expectations too. The bot should cut repeat questions, not create cleanup work. Every conversation should be visible in one place. And the cost should stay predictable as volume grows.

Grounding: answers from your content, or an honest "I don't know"

Ask every vendor two questions: where does the answer come from, and what happens when the source does not cover the question? SupportHQ answers only from the sources you approved: uploaded files, pages synced from your website, articles your team wrote, and past conversations you turned into knowledge.

When the answer is not there, the agent says so and brings in your team. Guardrails against made-up answers and prompt injection ship with the platform. You do not write them. See how to prevent AI hallucinations in customer support for the checks that matter.

Handoff: a person gets the conversation, with context

A chatbot without a good handoff traps customers. SupportHQ escalates on its own when confidence is low, when a customer asks for a person, or when the request needs an action it cannot take, such as a refund or an account change.

Each escalation carries the agent's reason, an urgency level, and a summary of the thread. Your team gets an alert by email, Telegram, or SMS. When a teammate replies from the inbox, the agent pauses on that conversation. Read how human handoff works.

Pricing model: what you actually pay for

AI chatbot pricing follows three common models. Per-seat pricing grows with your team. Per-resolution pricing grows with the bot's success, so a good month costs more. A flat monthly price with a conversation allowance stays predictable.

SupportHQ uses the flat model. Starter is $29 per month with 2,000 conversations. Growth is $49 per month with 3,000 conversations. There are no per-seat fees and no per-resolution fees. Compare the models in how much AI customer support costs.

Channels: answer where customers already ask

A customer support chatbot that lives only on your website misses the customers who message you elsewhere. The same SupportHQ agent, with the same knowledge base, runs in the website widget, in Telegram, and in Discord. WhatsApp is coming soon.

Group chats need a judgment a website widget does not. You write plain-English rules per channel that decide when the agent replies and when it stays quiet. The AI support agent page covers that behavior in detail.

Measurement: see why it handed off

Every conversation, from every channel, lands in one unified inbox with full history. The overview lists recent handoffs with the agent's reason and summary. Those reasons show what to add to the knowledge base next.

When your team answers a new question well, turn that conversation into a knowledge base article. The next customer who asks gets the same answer from the agent, and your escalation rate drops over time.

What SupportHQ does not do

Knowing the limits helps you decide faster.

It does not use decision trees. There are no flows to draw or maintain.
It does not take actions in your other systems, such as issuing a refund or changing an account. It escalates those requests to your team.
It does not answer from the open internet. If your content does not cover a question, it says so.
It does not run on WhatsApp yet. WhatsApp is coming soon.

FAQ

What is a customer service chatbot?

A customer service chatbot answers customer questions in a chat window or a messaging app, without a person typing each reply. Older bots follow scripted decision trees. Newer ones, like SupportHQ, write answers from your own documentation and hand off to a person when they cannot help. SupportHQ works as an AI agent: it decides when to answer, when to stay quiet, and when to escalate.

Is an AI chatbot for customer service better than a rule-based one?

For open questions about a product, yes, as long as it is grounded in your content. A rule-based bot is predictable but fails on questions nobody planned for. An AI bot handles new phrasing, and it stays accurate only when it answers from sources you approved.

How do I train the chatbot on my content?

Upload files, sync pages from your website, or write articles in the dashboard. You can also turn conversations your team answered into knowledge. There is no separate training project and no flow to draw.

What happens when the chatbot cannot answer?

It says it does not know. Then it escalates to your team with a reason, an urgency, and a summary of the conversation. Your team gets an alert by email, Telegram, or SMS and replies from the inbox.

Does it answer in languages other than English?

Yes. It replies in the language the question arrives in, with no setup. Your knowledge base can stay in one language.

How is SupportHQ priced?

Starter is $29 per month with 2,000 conversations. Growth is $49 per month with 3,000 conversations. There are no per-seat fees and no per-resolution fees.

Can I try it before I pay?

Yes. Every plan starts with a 7-day trial. A card is required to start the trial. Cancel before the trial ends and you are not charged.