An AI chatbot for business answers customers on your website or WhatsApp from your own approved information and passes the rest to your team. With Praxen, a chatbot on one channel is built as a 2–4 week pilot for £2,900 (£2,030 under the launch offer), with optional care at £350 a month (£245) and AI usage billed at provider cost to your own accounts; prices exclude VAT.
The parts of a chatbot price
Whoever writes the quote, a chatbot price is made of the same few parts. Looking at them one by one is the easiest way to compare two quotes.
- Setup. Collecting and checking your content, building the knowledge base, connecting the website widget or WhatsApp number, writing handover rules and testing on real questions.
- Monthly care. Updating answers when prices or policies change, reviewing conversations, correcting wrong answers and watching for errors.
- AI usage. The model provider charges for the text each conversation processes, so the bill grows with the number and length of conversations.
- Channel fees. Meta charges for some WhatsApp messages, mainly templates your business sends first, such as reminders, at rates that vary by country.
- Tools you already pay for. A helpdesk such as Zendesk or Intercom, or an online booking system, keeps its own subscription.
What a chatbot costs with Praxen
A chatbot on one channel is built as our implementation pilot. A chatbot on several channels, or one that works across your systems to look up orders and start returns, is a custom solution. The table shows launch offer prices; the standard ones are £2,900, from £6,500 and £350 a month. VAT is added on top.
What pushes the price up or down
- Number of channels. Each one needs its own connection and tests, and WhatsApp also needs a business account on Meta’s platform.
- State of your content. A current help centre and price list load quickly. Scattered PDFs and old email threads need sorting first.
- Answers or actions. A chatbot that only answers is the simpler build. Booking, looking up orders or starting returns means connecting to your systems with access rules and approval points.
- Integrations. A standard calendar or online shop connects faster than a system with custom fields or no API.
- Languages. Every extra language gets its own set of test questions.
- Handover. Passing conversations into Zendesk, Intercom or Freshdesk with a summary takes more work than an email alert.
- Conversation volume. More conversations mean a larger AI usage bill and possibly more WhatsApp charges. Our setup and care fees stay the same.
An illustrative payback check
The figures here are an example, only to show the sum. Say your team answers 25 repeat questions a day across chat, email and WhatsApp, and each one takes four minutes. Over 21 working days that is about 35 hours a month. At £18 an hour, those questions cost around £630 a month in staff time.
If the chatbot answers most of them from your knowledge base, set that figure against care at £245 a month plus AI usage. The build fee is paid once. Put your own volumes into the sum; the free 30-minute audit begins with them.
How to compare chatbot quotes
Send each supplier the same questions and put the answers side by side.
| Ask the supplier | A clear answer | Probe further if |
|---|---|---|
| Is AI usage included, and at what rate? | Usage at cost on your account, or a stated amount with a stated limit | It is “included” with no limit given, or marked up with no figure |
| Who owns the content and the accounts? | You do, and you can export everything | The knowledge base lives only on the supplier’s platform |
| How is the monthly fee charged? | A flat fee for agreed work | Per conversation or per resolution, with no estimate for your volume |
| What is tested before customers see it? | A test set from your real questions and a pass mark you agree | “We test as we go” |
| How does handover work? | Into your helpdesk or inbox, with the conversation and a summary | There is no route to a person |
| What happens to our data? | An enterprise API that does not train on it, and a data processing agreement | No written answer |
| How long is the contract? | Monthly care you can stop | A long minimum term before you have seen results |
Red flags in a chatbot quote
- No setup work described. Someone still has to gather and check your content, and if the quote leaves it out, the job falls to your team.
- A promise that it will answer everything. A good chatbot says when it does not know and hands the customer to a person.
- A handover that loses the conversation, so the customer has to start again with your team.
- AI usage folded into a flat fee with a cap nobody mentions.
- No test before customers see it, and no agreed result that counts as a pass.
- Content locked in the supplier’s platform with no way to export it.
- Prices the chatbot may quote with no single approved source behind them.
How to get a quote you can compare
- Pull a month of customer questions from chat, email and WhatsApp and sort them by topic. The largest groups set the scope.
- List the channels you need now and the ones that can wait.
- Name the systems the chatbot should read from, such as your booking system or online shop.
- Decide who receives handovers and in which tool.
- Send the same brief to every supplier, so each quote covers the same work.