On this page
The honest answer is that an AI chatbot can cost anywhere from nothing to several thousand dollars, and the gap between those two numbers is not mostly about quality. It is about what the chatbot is allowed to do.
A bot that answers questions from your website copy is cheap, because answering is cheap. A bot that checks live availability, books the appointment, writes the customer into your CRM and escalates the awkward conversations costs more — each of those is a connection to a real system somebody has to build and maintain.
This guide breaks the cost into the five parts every quote is made of — platform, setup, AI usage, integrations and maintenance — using pricing read from providers' own pages in August 2026. Every figure is linked at the end.
AI chatbot cost at a glance
Self-service platforms start free and run to a few hundred dollars a month. Professionally implemented chatbots typically combine a one-time setup fee with a monthly plan. The single biggest price driver is not conversation volume — it is how many of your business systems the chatbot has to touch.
How much does an AI chatbot cost in 2026?
There is no single price because a chatbot is not a single product — two things sold under the same name can differ as much as a printed FAQ sheet differs from a receptionist. What you are buying is some combination of five things:
- The platform — the software the chatbot runs on. Subscription, usually monthly.
- The implementation — getting your knowledge into it, designing the conversation, connecting it to your systems, testing it. One-time, and the part most often underestimated.
- AI usage — what it costs to actually run the model behind each conversation. Sometimes bundled, sometimes billed separately.
- Integrations — every system the chatbot reads from or writes to.
- Maintenance — keeping the answers current as your business changes.
Providers bundle these differently, which is why quotes are hard to compare: one vendor's $150 a month includes everything, another's excludes AI usage and meters it on top. The buyer questions further down exist to make that comparison possible.
AI chatbot pricing by type
Rule-based chatbots
The oldest kind: a decision tree with buttons. The visitor picks from options you wrote in advance, and the bot follows the branch. There is no language model involved, so there is no AI usage cost at all.
These are cheap and often free — HubSpot includes a chatbot builder in its free tier. They work when your questions genuinely are a short, fixed list, and fail the moment a visitor phrases something you did not anticipate. Most businesses outgrow one within months.
Self-service AI chatbots
Subscription platforms where you upload your own content — pages, PDFs, help articles — and the bot answers from it. The largest category, and where most small businesses start. Published pricing, read from each provider's page in August 2026:
Watch the unit, not the price
Chatbase counts message credits; Tidio counts billable conversations; HubSpot counts seats. A conversation might be twenty messages. These numbers are not comparable until you convert them to the same unit using your own expected volume — and that conversion is where a cheap-looking plan often stops being cheap.
Professionally implemented AI chatbots
Here the software is the smaller part of the cost. You are paying for someone to decide what the chatbot should do, prepare your content, connect it to your calendar and CRM, test it and keep it working. The structure is a one-time setup fee plus a monthly plan — what that setup covers is below.
Enterprise AI chatbots
At the top of the market, pricing shifts to per-seat or per-outcome models. Intercom publishes seats at $29, $85 and $132 per seat per month, with its Fin AI agent from $0.99 per resolution on top. Zendesk publishes Suite Team at $55 and Suite Professional at $115 per agent per month billed annually, and bills AI agents per automated resolution at a rate it does not publish.
Per-resolution pricing is appealing in principle — you pay when the bot actually solves something. Model it at your real volume first: a thousand resolved conversations a month at $0.99 is roughly a thousand dollars, which is a different proposition from a flat subscription.
What does AI chatbot setup cost include?
This is the line item people query most, so it is worth being specific. A professional setup fee typically covers:
- Discovery — what the chatbot should handle, and just as importantly what it must never attempt.
- Knowledge preparation — gathering your pricing, policies, services and FAQs into something a model can answer from accurately. Usually the single largest chunk of the work.
- Conversation design — the greeting, the tone, how it asks for a name and email, when it escalates.
- System instructions — the written rules constraining what the bot can say and how it behaves when it does not know something.
- Integration work — connecting the calendar, the CRM, the helpdesk. Scales with the number of systems.
- Lead capture configuration — what gets collected, where it lands, who is notified.
- Installation and branding — getting it on the site looking like it belongs there.
- Testing — running real questions at it, including the awkward ones, and fixing what comes back wrong.
- Human handoff — the route to a person, and what happens outside business hours.
- Analytics — so you can see what people actually asked.
A $0 setup fee is not free
It means the work above is yours. That is a perfectly reasonable trade if you have the time and the content is already tidy — but budget the hours honestly. Knowledge preparation alone is usually a day or two of somebody's attention, and it is the step that determines whether the bot is accurate.
What does an AI chatbot cost per month?
Recurring cost breaks into a handful of components. Some vendors bundle all of them into one number; others itemise.
The two most often excluded are maintenance and monitoring. A chatbot is not static — prices change, services change, and an answer that was right in March is wrong by September. Somebody has to own that.
Do AI chatbots have usage costs?
Usually yes, in one form or another — either metered directly, or hidden inside a message or conversation cap.
In plain terms: every time your chatbot replies, it sends the conversation so far — plus relevant chunks of your knowledge base — to a language model, and the model sends words back. Both directions are billed in tokens, roughly three-quarters of a word each. The AI glossary defines tokens, embeddings and retrieval if you want the longer version.
OpenAI's published rates in August 2026 give the scale: its lightweight model at $0.20 per million input tokens and $1.20 per million output, its mid-tier model at $2.00 and $12.00.
A worked example — and its assumptions
Take a five-exchange support conversation. Assume roughly 10,000 input tokens total (system instructions plus retrieved knowledge plus the growing conversation history, resent each turn) and 1,000 output tokens. On the lightweight model that is about $0.003 — a third of a cent. On the mid-tier model, about $0.03. At 1,000 conversations a month: roughly $3 and $32 respectively. Those token counts are an illustration, not a measurement; a bot retrieving large documents will use considerably more.
The point is not the figure — it is that raw model cost is rarely what makes a chatbot expensive. When a vendor charges $150 a month for 4,000 messages, you are mostly paying for the platform, the retrieval system and the support around it. That is legitimate; it is just worth knowing which part is which.
How integrations affect AI chatbot pricing
This is the cost driver that surprises people, and the honest reason two chatbot quotes can differ by a factor of five. A chatbot that only answers questions needs one thing: your content. A chatbot that does things needs an authenticated, error-handled connection to every system involved — and each connection is a small software project with its own failure modes.
Common integrations — HubSpot, Salesforce, Zoho, GoHighLevel, Shopify, WooCommerce, Google Calendar, Calendly, Slack, email and mainstream helpdesks — are well-documented and reasonably quick. The expensive ones are older systems with no API, or in-house databases where it must be built from scratch. Ask early which category yours falls into; it changes the shape of the project.
Every integration also adds something that can break. Start with the one or two carrying real business value, and add more once those are boring.
What should a small business budget for an AI chatbot?
Four illustrative scenarios. These describe requirements, not quotes — what any particular provider charges depends on the systems involved.
Scenario 1 — Local service business
A plumber, a salon, a small clinic. Needs: hours, coverage area and pricing questions; out-of-hours lead capture; straightforward booking; handoff for anything unusual. One calendar integration, no CRM. The lightest real deployment, and often the fastest payback — the alternative is a missed enquiry.
Scenario 2 — E-commerce business
Needs: sizing questions before purchase, order status after it, returns eligibility, escalation when something is wrong. Requires a store integration and a read into order data — so a security conversation about what the bot can see. Volume is higher, so usage limits matter most here. Our e-commerce automation guide covers the wider picture.
Scenario 3 — Professional services firm
An accountancy practice, a law firm, an agency. Needs: qualify enquiries against the work you actually take, write leads into the CRM, book consultations, answer from a substantial body of material. More integration work, and more care over what the bot may say — the constraint here is professional, not technical.
Scenario 4 — Larger organisation
Multiple departments, several integrations, real volume, role-based permissions, reporting by team, a security review before launch. Here per-seat or per-resolution pricing becomes the relevant comparison — and procurement often costs more than the software.
Should you build an AI chatbot yourself?
For a real set of needs, yes. Self-service platforms have become good: a website bot answering from your existing content can be working in an afternoon.
The dividing line is not company size. It is whether the chatbot needs to touch other systems, and whether anyone has the hours to own it. A bot nobody maintains degrades quietly — it keeps quoting last year's prices, and nobody notices until a customer does.
The question is not whether you can build it. It is whether, three months from now, anyone will still be updating it.
AI chatbot costs businesses often overlook
- Content cleanup. If your pricing lives in three contradictory places, the bot will confidently pick one. Fixing that is the project's real first task.
- Testing time. Someone has to ask it two hundred awkward questions before customers do.
- Staff training. Your team needs to know what it handles, what it escalates, and how to correct it.
- Monitoring. Nobody watching means nobody notices when quality drifts.
- Knowledge updates. Every price change, new service and policy revision is a small maintenance task.
- Failed conversations. A bot that cannot answer still costs you — in the customer's patience, and in the human who picks up afterwards.
- Escalation capacity. Deflection only works if there is a person on the other end of the handoff.
- Data and compliance. Chat transcripts are personal data, with retention, storage-location and access obligations attached.
- Migration. Moving platforms means rebuilding the configuration. Ask who owns your prompts and conversation data before you need the answer.
Is a cheap AI chatbot good enough?
Often, yes — and anyone who tells you otherwise is selling something.
A cheap chatbot is sufficient when traffic is modest, the questions are simple and documented, nothing needs to connect to another system, and the bot is not standing between a customer and a purchase. There, a subscription platform and an afternoon of your time is the right answer; spending more is waste.
It changes when the chatbot sits on a revenue path. If it is qualifying leads and booking appointments, or handling support that would otherwise reach a person, the cost of it being wrong is no longer measured in the subscription fee. That is when implementation quality outweighs platform price.
How to calculate AI chatbot ROI
The framework is simple enough to do on paper in ten minutes.
The calculation
Monthly value = staff hours saved × hourly cost + additional leads captured × conversion rate × average customer value + support cost avoided − monthly chatbot cost. If the result is comfortably positive at conservative inputs, the case is there. If it needs optimistic inputs to work, it is not ready yet.
A hypothetical example. Suppose a local service business has someone spending 25 hours a month on repetitive website and email questions, at a loaded $25 an hour — $625 of staff time. The chatbot handles 60%, returning about $375. It also captures four after-hours enquiries that previously went nowhere, one of which converts at an average job value of $400.
That is roughly $775 of monthly value against a cost in the low hundreds. Illustrative figures showing the method — not a TensoraAI customer result. Run it with your own numbers and be pessimistic on the conversion rate; it is the input people overestimate most. Worth tracking once live:
- Conversations handled, and how many ended without a human
- Qualified leads captured, especially outside business hours
- Appointments booked
- First-response time
- Deflection rate — support contacts resolved without a person
- Conversion rate for visitors who used the chat versus those who did not
- Human support hours actually returned
Deflection alone is a misleading metric
It can be improved by making it hard to reach a human, which looks like success on a dashboard and shows up later as churn. Track deflection alongside satisfaction, or you will optimise the wrong thing.
AI chatbot pricing models compared
10 questions to ask an AI chatbot provider before you buy
Ask every provider the same ten. The differences in their answers will tell you more than the headline prices.
- What exactly is in the setup fee? Ask for it itemised. A real build and a template with your logo on it are different products.
- What is in the monthly fee — and what is not? Specifically: maintenance, monitoring, and changes.
- Is AI usage included? If not, what is the rate, and what does it cost at my expected volume?
- Are there message, conversation or seat limits? What happens when I exceed them — throttling, overage, or an upgrade?
- Which integrations are already built, as opposed to possible? "We can integrate with anything" is not an answer.
- Can it book appointments into my actual calendar? Live availability, or a request someone processes later? These are very different outcomes.
- How does handoff to a human work? Does the person receive the conversation, or does the customer start again?
- Where is my data stored, for how long, and who can access it? Get this in writing before launch, not after.
- How are wrong answers found and fixed? Who reads the transcripts, and how often?
- What happens if I leave? Who owns the prompts, the knowledge base and the conversation history, and can I export them?
The one that reveals the most
Question ten. A provider comfortable telling you exactly how you would leave is usually a provider not relying on it being difficult.
How TensoraAI approaches AI chatbot pricing
We publish our numbers rather than hiding them behind a discovery call. The alternative mostly exists to price-discriminate, and it wastes a week of your time before you learn whether the range is even plausible.
The model is a one-time setup fee plus a monthly plan. Setup covers the build — knowledge preparation, conversation design, integrations, testing and installation. The monthly plan covers hosting, support and keeping it accurate as your business changes. Chatbot plans run from a Starter tier for a fixed set of questions, through a Growth tier adding appointment booking with calendar and CRM integration, to an Enterprise tier with multiple languages and custom API work.
Current figures for all three tiers, and the ongoing support plans, live on the pricing page — deliberately the only place they live. How we build these is on the AI chatbot service page; what we have actually implemented is in our case studies.
What we will not do is quote you a setup fee for work that does not need doing. If your requirement is a website bot answering from your existing content with no integrations, a self-service platform is a better use of your money, and we will say so.
Frequently asked questions
How much does an AI chatbot cost?
Anywhere from free to several thousand dollars, depending on what it does. Self-service platforms published entry plans between roughly $24 and $40 a month in August 2026, with higher tiers at $150 and above. Professionally implemented chatbots usually combine a one-time setup fee with a monthly plan.
How much does an AI chatbot cost per month?
Most small businesses on a self-service platform land between $25 and $150 a month depending on volume. For a managed implementation the monthly plan covers hosting, support and maintenance alongside a setup fee. Always check whether AI usage is included in that figure or metered separately.
Is ChatGPT free to use for a business chatbot?
The consumer ChatGPT product is not a website chatbot — it has no connection to your site, content or systems. Building on OpenAI's API means paying per token, which is cheap per conversation but is only the model cost. You still need the interface, the retrieval system, hosting and maintenance.
Why do custom AI chatbots cost more?
Because they do more than answer. Checking live availability, writing to a CRM, looking up an order and handing off to a human are each a working connection to another system, with authentication, error handling and a security review attached.
Are AI API costs included in chatbot pricing?
Sometimes. Platforms selling message credits or conversation allowances have usually priced the model cost in. Providers building on your own API keys may bill it through. Ask what the rate is once the included allowance runs out — that is where surprise bills come from.
How much should a small business spend on a chatbot?
Work backwards from value, not from a budget. Price the staff hours going into repetitive questions and the enquiries missed outside business hours. If a chatbot plausibly recovers more than it costs at conservative assumptions, it is worth doing.
Can I build an AI chatbot myself?
Yes, and for simple needs you should — self-service platforms can have something answering from your content within a day. The honest limits are integrations, testing and ongoing maintenance. A bot nobody updates keeps confidently quoting last year's prices.
Is an AI chatbot worth the investment?
It depends on whether visitors currently go unanswered. A business with steady web traffic and repetitive enquiries usually sees the case quickly. One with few visitors, or whose enquiries are all complex and high-value, often does better investing in the phone or back-office automation instead.
Does an AI chatbot need ongoing maintenance?
Yes, and it is the most underestimated cost. Prices, services and policies change, and each change makes an existing answer wrong. Somebody has to read the transcripts, spot the gaps and update the knowledge base — either you, or the provider as part of a monthly plan.
How long does it take to build an AI chatbot?
A self-service bot answering from existing content can be live the same day. A professional implementation usually takes days to a few weeks, and the timeline is driven by integration work rather than conversation design — a modern cloud calendar is quick, an older system with no API is a project.
If you want the wider case for and against — benefits, deployment patterns and where chatbots fail — start with our guide to AI chatbots for small business. If the phone is a bigger problem for you than the website, how much an AI voice agent costs runs the same analysis for voice. And if you are not sure which of the two matters more, that is exactly what the free AI audit is for.
Sources and pricing references
Every third-party figure in this article was read from the provider's own pricing page in August 2026. SaaS pricing changes frequently — check the source before relying on a number, and treat anything unlinked as an illustration rather than a quote.
- Chatbase pricing — plan tiers and message credits.
- Tidio pricing — plan tiers, billable conversations and the Lyro AI add-on.
- HubSpot pricing — free chatbot builder and paid tier entry points.
- Intercom pricing — per-seat plans and Fin per-resolution pricing.
- Zendesk pricing — Suite per-agent plans and the automated-resolution billing model.
- OpenAI API pricing — per-million-token input and output rates used in the worked example.
- TensoraAI's own setup fees and monthly plans come from the pricing page, which is the source of truth for them.
Botpress and Voiceflow are frequently recommended in this category but do not publish plan pricing on their public pricing pages, so no figures for either appear above. That is worth knowing before you shortlist them — unpublished pricing means a sales conversation before you can compare.
