How much does it cost to add AI to your website?

Adding AI to a site you already have is rarely one big invoice. It is a stack of small decisions — which feature, off-the-shelf or custom, a flat monthly fee or pay-per-use — and each one moves the number a lot. This guide puts real 2026 figures against the common options so you can budget before you talk to anyone.

The four AI features people actually add

Most requests fall into one of four buckets, and they price very differently:

• A chatbot / support agent — answers FAQs, deflects tickets, qualifies leads.
• AI site search — natural-language search that understands intent, not just keywords.
• Recommendations — "you might also like" and personalised product surfacing.
• Content generation — drafting descriptions, meta text, or blog copy from your data.

Chatbots are by far the most common first step, so they anchor most of the pricing below. If you are weighing which features are worth it in the first place, our companion piece on AI features that increase sales is the better place to start; this post is about the money.

Off-the-shelf: the cheapest way in

For a hosted, plug-and-play chatbot you are mostly paying a monthly subscription and doing the setup yourself.

• Entry tiers cluster around $19–$50/month. Zendesk, Freshchat and LiveChat all start near $19 per seat; Tidio's Starter is about $24/month for ~100 conversations [Tidio].
• Website add-on tools have pushed prices down — SiteGPT's chatbot add-on sits at $39/month, while white-label options like Chatbase run around $199/month [Tidio, Crescendo].
• Growing teams typically land at $60–$200/month once they need more conversations or seats — Tidio's Growth plan is roughly $180/month for ~1,000 conversations, and Intercom's Advanced seat plan starts near $170/month [Tidio].

For AI site search, Algolia is a useful benchmark: its free tier includes 10,000 searches/month, then charges about $0.50 per 1,000 additional searches on the Grow plan, rising to $0.75–$1.75 per 1,000 once AI features are switched on. A site doing 100,000 searches a month pays roughly $45/month in search overage alone [Algolia].

Sources: Tidio, Crescendo, Algolia pricing.

Pay-per-use is the pricing model to understand

The biggest shift in 2026 is away from flat per-seat fees toward per-resolution (or per-outcome) billing — you pay when the AI actually resolves something.

The reference point everyone quotes is Intercom's Fin AI agent at $0.99 per resolution, with a minimum of 50 resolutions a month [Gleap, Featurebase]. That headline looks cheap, but two details decide your real bill:

• Resolution rate. In practice, Fin resolves 42–50% of conversations on its own [Gleap]. So if 1,000 people ask questions, you might pay for ~450–500 resolutions — roughly $450–$500 that month — while the rest still reach a human.
• Add-ons stack. Copilot for your agents, proactive messaging and analytics tiers each carry their own fee, often $29–$99/month on top [Featurebase].

The lesson: model your own volume before signing. Plug your monthly conversations and expected resolution rate into our chatbot ROI calculator to see whether pay-per-use or a flat plan is cheaper for you, and how quickly it pays back.

Sources: Gleap, Featurebase.

Custom builds: when the ready-made option is not enough

If you need the AI to plug into your CRM, booking system or internal database — or to behave in a way no template allows — you are into custom development. This is a one-off build cost plus ongoing running costs.

• Basic FAQ bot: from about $5,000 [Crescendo, Easycomm].
• Mid-market, NLP-powered with CRM integrations: most businesses invest $15,000–$35,000 [Crescendo].
• Enterprise / agentic systems: $50,000–$200,000+, depending on integrations and scale [Biz4Group, Crescendo].

Then budget for upkeep. A custom bot needs retraining, monitoring and updates — commonly $500–$3,000/month [Crescendo]. That figure is easy to forget, and it is exactly where custom projects go over budget.

Sources: Crescendo, Easycomm, Biz4Group.

Setup vs monthly: read the whole invoice

Two costs get overlooked:

• Setup and integration. Most SaaS tools advertise "no setup fee", but connecting them to your data, training them on your content and wiring up WhatsApp or SMS via the Business API all take time — yours or an agency's.
• Total cost of ownership. One audit of 12-month chatbot spend found actual costs averaged 2.3x the listed subscription once usage overages, integrations and support were counted [Crescendo]. Always budget above the sticker price.

A sensible rule of thumb: the subscription is the floor, not the ceiling.

What actually pays off

Money is easiest to justify where AI removes repetitive work or recovers lost revenue:

• Support deflection. If a chatbot resolves even 40% of routine questions, it frees your team for the conversations that need a human — the clearest, fastest payback.
• Search and recommendations on stores. Better product discovery lifts conversion, so overage-based search pricing usually pays for itself on any site with real traffic.
• Content generation saves hours but rarely drives revenue directly — treat it as an efficiency win, not a growth lever.

Start with the one feature that touches your biggest bottleneck, prove the return, then expand. For a fuller breakdown of ongoing running costs specifically, see our pillar guide on what an AI chatbot costs.

Where to start

For most SMBs the honest answer is: $20–$200 a month gets you a capable off-the-shelf feature, pay-per-use suits unpredictable volume, and a custom build only makes sense once a ready-made tool genuinely cannot do the job. Run your numbers through the chatbot ROI calculator first — it turns "is this worth it?" into a figure.

If you would rather not sort through platforms alone, see the AI tools we build and integrate, or book a free consultation and we will help you pick the cheapest option that actually solves your problem.