AI Chatbot Cost for Small Business (2026 Prices) | TEC

An AI chatbot costs a small business $0 to $400 a month on a no-code platform, $2,000 to $25,000 for an agency-built bot, or $10,000 to $150,000+ for fully custom development. The right band depends on your chat volume, what the bot must do beyond FAQs, and whether you build or hire. Prices below were checked in October 2026.

Key takeaways

• Most small businesses pay $29 to $150 a month for a no-code AI chatbot, plus usage charges that depend on chat volume.

• Chatbots are billed in three incompatible units: per message credit, per AI conversation, or per resolution. Decode the unit before comparing prices, or a cheap-looking plan becomes an expensive bill.

• Watch the dual-meter trap: per-seat base fees plus per-outcome charges stack. A two-seat Intercom setup resolving 500 chats a month can cost over $550.

• The free plan is enough until the bot is handling real customer volume. If it is booking appointments or recovering carts at night, the paid plan has already paid for itself.

• Hiring a developer makes sense when the bot must talk to your CRM, ordering system, or booking tool. DIY covers FAQ answering and lead capture.

What does an AI chatbot cost in 2026?

Start with the three ways to get one, because they are really three different purchases. A pricing review of build routes read on October 6, 2026 puts the bands at:

Route by route: a DIY no-code platform costs $0 to set up, $0 to $400 a month, and goes live in hours to days (first year: $0 to $4,800). An agency-built bot costs $2,000 to $25,000 to set up, $500 to $5,000 a month, and takes 2 to 8 weeks (first year: $8,000 to $85,000). A fully custom build costs $10,000 to $150,000+ to set up, $500 to $5,000 a month in running costs, and takes 3 to 9 months (first year: $16,000 to $210,000+).

The DIY route is the right starting point for most small businesses. You train the bot on your own website, help docs, and FAQs, and it answers customers on your site. The middle route makes sense when you need the bot wired into your booking system, CRM, or store. The custom route is for complex builds with many integrations or regulated data, and honestly, most SMBs never need it.

On the custom-build side, a cost analysis built from 2026 vendor datasets puts the median of published quotes at $14,000 to $45,000 for one AI agent, with proof-of-concept work around $8,900 to $10,000 and multi-agent systems starting near $50,000. Agency hourly rates for chatbot developers run from under $25 to around $200 an hour, with $50 to $99 the most common band, according to Clutch's 2026 data cited in an analysis updated in September 2026.

How do chatbot platforms actually charge?

This is where small business owners get burned, because the headline price is almost never the whole bill. Chatbot platforms use three billing units that do not convert neatly into each other:

• Per message credit. Chatbase charges this way: the Hobby plan is $40 a month for 700 message credits, Standard is $150 for 4,000. One chat conversation usually burns several credits.

• Per AI conversation. Tidio's Lyro add-on works this way: from $39 a month for 50 AI conversations (about $32.50 a month billed annually), on top of a base plan that starts at $29 a month. Every account also gets 50 Lyro conversations free, once.

• Per resolution or outcome. Intercom's Fin agent bills $0.99 per resolved outcome, on top of seat plans that start at $29 per seat a month. Zendesk charges $1.50 per automated resolution on a committed plan ($2.00 pay as you go), plus seats from $55 an agent a month. ChatBot.com's 2026 plans start at $19 a user a month billed annually, including 10 AI resolutions.

So the pricing looks like this at the entry level, from comparisons that checked each vendor's pricing page in early October 2026:

In short: Chatbase from $40 a month (message credits), Tidio $29 a month base plus Lyro from $39 a month (per AI conversation), Intercom $29 a seat a month plus $0.99 per outcome, ChatBot.com from $19 a user a month on annual billing, Zendesk from $55 an agent a month plus $1.50 per resolution, and Copilot Studio $200 a month for 25,000 credits.

Watch the dual-meter trap. Two meters running at once, seats and outcomes, can double or triple the headline price. This is the number-one reason buyers feel overcharged. The worked examples in the next section show exactly how it stacks up.

What would you actually pay at your chat volume?

Nobody buys "a chatbot." They buy a bill for their chat volume. Here are worked monthly estimates using the October 2026 prices above. Assume two seats where seat plans apply.

At 100 AI-handled chats a month:

• Tidio: $29 base + about $39 Lyro = roughly $68

• Chatbase: $40 Hobby (700 credits covers this easily) = $40

• Intercom: $58 for two seats + 100 x $0.99 outcomes = $157

At 500 AI-handled chats a month:

• Tidio: $29 base + Lyro scaled to volume (bigger conversation packs) = roughly $100 to $150

• Chatbase: $150 Standard (4,000 credits) = $150

• Intercom: $58 seats + 500 x $0.99 = $553

• Zendesk: $110 for two seats + 500 x $1.50 = $860

• ChatBot.com: $38 for two users on Essential + overage resolutions beyond the included 10 = roughly $100+

At 3,000 AI-handled chats a month, the per-resolution models get expensive fast: Intercom runs past $3,000 before seats, while flat-credit platforms like Chatbase's Pro ($500) or a Copilot Studio pack ($200 for 25,000 credits) can be an order of magnitude cheaper. This is why the billing unit matters more than the headline price. High volume plus per-resolution pricing is the most expensive combination in the market.

A note on what is excluded: these are software bills only. They do not include setup help, training the bot on messy documents, or integrations. Those are the agency and custom costs from the first table.

What pushes the price up or down?

Five factors move almost every quote. Run through them before you sign anything:

1. Knowledge base size and messiness. Answering from 50 clean FAQ pages is cheap. Reasoning across thousands of PDFs, spreadsheets, and legacy wikis is an engineering project, because ingestion, chunking, and retrieval tuning become real work.

2. Integrations. Each connection to your CRM, booking system, ordering platform, or helpdesk adds complexity. Simple webhook integrations can take a day. Legacy systems can take weeks, and agencies price each integration in the low thousands.

3. Chat volume and peaks. Weekend rushes, seasonal spikes, and growth all hit the usage meter. Price for the busy month, not the average one.

4. The AI model doing the work. Raw API costs vary widely by model. Platform pricing hides this from you, but custom builds pass it through: routing simple questions to a cheap model and hard ones to a capable model can cut running costs by half or more.

5. Channels and branding. A bot on your website only is cheapest. Adding WhatsApp, Instagram, or voice costs more. Removing the vendor's branding usually costs extra too, like a $50-a-month add-on on some platforms.

If a quote looks surprisingly low, check which of these the seller assumed away. That is where the surprise invoices live.

Is an AI chatbot worth it for a small business?

Usually, yes, if customers actually message you. Salesforce's State of Service: AI Agents Edition, a survey of 3,075 service professionals published on May 20, 2026, found that 66% of service organizations now use agentic AI, up from 39% the year before. Seventy percent of adopters saw measurable value within 60 days, and the number-one improved KPI was customer satisfaction, ahead of productivity and handle time. That is the honest value case: fewer missed questions, faster answers, happier customers.

The math is simple at small scale. One part-time support hire costs $1,500 to $2,500 a month. A $40 to $150 platform bot that deflects even a third of routine questions pays for itself many times over, and it answers at 11 p.m. when nobody is on shift.

There are cases where it is not worth it. If you get fewer than a couple dozen customer messages a month, a shared inbox and good templates do the job for free. If your conversations are legally sensitive or emotionally charged, a bot handing off badly costs more trust than it saves in wages. And a bot nobody maintains goes stale: someone has to review its answers monthly, or it starts answering from last year's prices.

The upgrade rule is one question: is the free plan costing you customers? If the bot is hitting usage caps during the hours you are closed, or it cannot take bookings and payments, the paid plan has already paid for itself. Our guide to AI receptionists covers the after-hours side of this decision in more detail.

How long does it take to get a chatbot live?

Typical timelines: DIY no-code platform, a few hours to a few days (the work is training it on your content and testing answers). Agency build, 2 to 8 weeks (conversation design, integrations, testing edge cases). Custom build, 3 to 9 months (architecture, knowledge-base engineering, security review).

The DIY number assumes your content is in decent shape. If your help docs contradict each other, plan a week to clean them up first. Garbage in, bot answers garbage.

Should you build it yourself or hire someone to do it?

Build it yourself if your needs fit this list: a few hundred chats a month or less, mostly FAQ-style questions, one channel (your website), and no deep integrations. A no-code platform, an afternoon, and a spreadsheet of your 30 most common questions will get you live.

Hire someone when the bot has to do real work: talk to your CRM or booking system, handle multi-step workflows like quotes and scheduling, work across WhatsApp and your site, or handle data you cannot afford to leak. That is integration and engineering work, and it is also where most DIY projects stall for six months and then get rebuilt anyway.

Deciding between a freelancer and an agency for the build is its own decision. Either way, vet the person like your customer experience depends on it, because it does. These are the red flags to watch when hiring an AI expert, and if you have not decided what the bot should even do yet, start with figuring out which part of your business to automate first.

How TEC handles this

TEC is a managed marketplace where businesses hire vetted experts, including AI specialists who build and deploy chatbots. Here is how hiring through TEC works, from the site's own how-it-works and policy pages:

• Post a brief (about 60 seconds). TEC typically shares a shortlist of 2 to 3 vetted experts within about 48 hours of the brief.

• Every expert goes through a 5-step screening that includes ID verification, portfolio review, and a skills or technical assessment. Matching considers skills, domain fit, past outcomes, and rate. No bidding.

• Client funds are held in escrow and released per milestone only after the client approves. If the client does not respond within 7 days of a milestone delivery, funds release automatically.

• A TEC project manager oversees milestones, scope, deadlines, and quality.

• Clients can also book a free advisor call to talk through what they need before committing.

Pricing depends on scope and the expert, so TEC never quotes price ranges up front.

To talk to someone who builds these for a living, hire a vetted AI expert through TEC: post your project free and get a shortlist with a full price breakdown in about 48 hours.

The cheapest chatbot is the one you actually use

Most small businesses overthink this decision. Start with a $0 to $40 plan, train it on your real customer questions, and run it for a month. You will learn more about what you need from four weeks of real chats than from four weeks of comparing pricing pages. Then either scale the plan up or bring in an expert to build the real thing.

If you want the real thing built right the first time, hire a vetted AI agent specialist or post your project on TEC and get a shortlist and price breakdown in about 48 hours. Free to post, no commitment.