Last updated
The chatbot setup tax is the time, decisions and extra bills between signing up and having an AI agent that helps customers. Here is where it comes from.
Picture a founder of a small software company who decides it's time for an AI support agent. The demo looked easy. A few weeks later they're still in the settings, mapping intents, writing sample conversations, sitting in onboarding calls and trying to decide which AI model should answer billing questions. Customers are still waiting on email in the meantime.
That gap is what we call the chatbot setup tax. It's all the time, decisions and extra invoices that sit between signing up for a tool and having an agent that actually helps customers. Some of that work is necessary. A lot of it isn't, and it tends to hit small teams hardest, because they don't have a project manager to absorb it.
Where the chatbot setup tax comes from
Many chatbot products were designed before language models could read a help centre and answer sensibly. They were built around intents and flows. You defined what a customer might want ("cancel my plan"), wrote the phrases they might use, and drew the path the bot should follow. Plenty of platforms have added AI since, but the setup screens often still carry that structure, so you end up doing flow-building work on a tool that no longer strictly needs it.
Large helpdesk suites add another layer. They do a lot, and for a big support organisation with many teams, workflows and integrations, a proper implementation project makes sense. For a business with a website, a few documents and a WhatsApp number, the same process can feel like buying a lorry to pop to the shops.
Then there's the stuff every setup has, whichever tool you pick. Your content has to be in reasonable shape. Someone has to decide when a person takes over. And the agent needs testing with real questions before customers see it. That part is worth the time. The rest is the tax.
Why choosing an AI model shouldn't be your job
Some platforms ask you to pick the language model during setup, or let you switch between several. That sounds like flexibility. In practice, most business owners have no good way to judge which model handles refund questions better, how the choice affects cost, or what changes when a vendor releases a new version.
We think that decision belongs to the platform. The people building the product test the models all the time and are in a much better position to choose. At SimplyBoost, answers are generated with Google Gemini, and you never have to pick or tune a model. You spend your time on the things only you know, like your prices, your policies and how you want to sound.
The extra invoices that come with setup
The second part of the tax is financial. Some tools ask you to connect your own AI provider account, which means a second bill that moves with usage. Others sell the AI as an add-on to a base plan. Tidio's Lyro AI agent, for example, starts at $32.50 a month for 50 conversations on top of a Tidio plan, and Zendesk's Copilot is $50 per agent a month on top of seats. None of that is wrong, but it does make the total harder to predict. We go through the different pricing models in more detail in our article on hidden AI chatbot pricing.
Our view is that the platform should carry the AI cost and give you one price. SimplyBoost plans are flat, from €39 a month for 500 AI replies, with no per-seat or per-resolution fees and nothing extra to pay an AI provider.
What a short chatbot setup looks like
Here's how it works in SimplyBoost, as an example of a setup that stays small.
- Add your knowledge. Paste your website URL (one page, or a crawl of up to 50 sub-pages), add a sitemap, or upload PDFs, Word files and FAQs. Pages and documents are processed in the background while you carry on.
- Set the basics. Choose one of eight tones, a reply length, the reply language, and write any rules or brand voice notes in plain text.
- Check the handover. Sensible triggers are already there, like a customer asking for a person, complaints, billing and refund questions, and two failed answers in a row. You can add your own keywords and choose how your team gets alerted.
- Go live where your customers are. Installing the website widget takes a few minutes with a script snippet, the WordPress plugin, the Shopify theme app embed or Google Tag Manager. WhatsApp connects through Meta's official sign-up flow, and Instagram connects through your Meta account.
There are no intents to map and no flows to draw. The agent answers only from the sources you added, and when it doesn't find a good match it says what it can help with and offers your team instead of guessing. Questions it couldn't answer collect in a Missing tab, and answering one there adds it to the knowledge base.
The part of setup that's worth your time
A short setup doesn't mean no effort. The agent can only be as good as what you give it, so it's worth checking that your pricing, delivery and return policies and opening hours are written down somewhere clear. Spend half an hour asking it the questions your customers really send, including the awkward ones. Then read the first week of conversations and fill the gaps that show up. That's the work that makes a difference, and no platform can do it for you.
Frequently asked questions
How long should AI chatbot setup take in 2026?
For a small business with a website and a few documents, the technical part can be short. Installing a widget takes a few minutes, and content is processed in the background. Leave time for testing with real questions and for reviewing the first conversations. Custom integrations and complex workflows take longer, and for a large support team that's often justified.
Should I choose which AI model my chatbot uses?
Ideally not. Comparing models well takes a lot of testing, and the right answer changes as new versions come out. A good platform makes that choice for you and keeps it up to date. SimplyBoost uses Google Gemini, so there's nothing to configure.
Why do some chatbot implementations take so long?
Usually a mix of reasons: setup screens built around intents and flows, many integrations, several teams and approval steps, and content that needs cleaning up first. Large helpdesk platforms serve big organisations with complex needs, so their onboarding reflects that.
What is a reasonable monthly chatbot bill for a small business?
It depends on your volume and what you need, so work out the total for your own numbers, including seats, AI usage and add-ons. SimplyBoost costs €39, €89 or €169 a month for 500, 2,000 or 6,000 AI replies, with lower prices on yearly billing. When a bill has several parts that move with usage, it's harder to predict in a busy month.
Can I switch chatbot platforms once I have set one up?
Yes, though how easy it is depends on the setup. If your agent learns from your website and documents, you can point a new tool at the same sources. Hand-built intents and flows rarely move across, which is one more reason to keep setup light.
A note on where we stand: we build SimplyBoost, a flat-priced AI support agent for websites, WhatsApp and Instagram. We're a Dutch company (KVK 87456346), hosted in Europe, which is why we care about keeping setup short. If you want to see for yourself, try it free for 7 days, no credit card needed.