The word "chatbot" covers two very different realities. Confuse the two and you risk overpaying for a gimmick, or disappointing your visitors with a bot that answers beside the point. This article explains what really sets a scripted chatbot apart from an AI chatbot, what each one costs, and how to choose the one that truly fits your business.
Rule-based and AI chatbots: two technologies, two promises
On paper, both display the same little bubble in the bottom-right corner of your website. In practice, the machinery running behind them has nothing in common. One follows a script written in advance, the other understands language and composes its own reply. Understanding this difference is the first step before you sign a quote or launch an in-house project.
The scripted chatbot: the bot that follows a script
It follows a programmed decision tree: buttons, menus, predefined answers. The user clicks "Track my order," the bot asks for a number, then returns a status. Reliable and predictable, but rigid: it only understands what was planned for. Step off the marked path and it invariably replies, "Sorry, I didn't understand your request."
- Upside: full control, no risk of an off-topic or inaccurate answer
- Upside: low cost and quick setup, often within a few days
- Upside: perfect for guiding, qualifying, or collecting a simple piece of information
- Limit: frustrating as soon as the user phrases their question in their own words
- Limit: every new case to cover means manually adding another branch to the script
The AI chatbot (LLM): the advisor that understands
Powered by a large language model (LLM), it understands natural language and produces written answers. Connected to your data through an approach like RAG, which plugs the AI into your own documents, it becomes a genuine advisor able to cite your catalog, your opening hours, or your return policy. The user can ask the same question ten different ways, and the bot keeps up.
- Upside: understands freely phrased questions, typos included
- Upside: answers with your up-to-date information, with no decision tree to maintain
- Upside: handles several topics within a single conversation, just as a human would
- Limit: requires serious guardrails to avoid inaccurate or made-up answers
- Limit: higher setup and usage costs, with a share of ongoing maintenance
A simple rule of thumb
If your visitors always ask the same three or four tightly scoped questions, a scripted chatbot is enough. If your products, services, or procedures are varied and customers express themselves freely, AI truly comes into its own.
Detailed comparison: scripted vs AI
| Criterion | Scripted chatbot | AI chatbot (LLM) |
|---|---|---|
| Flexibility of understanding | Low | High |
| Setup cost | $ | $$ |
| Monthly running cost | Fixed and low | Varies with volume |
| Risk of error | Near zero | Manageable with guardrails |
| Maintenance | Manually adding branches | Updating the knowledge base |
| Wow factor | Low | High |
| Time to go live | A few days | A few weeks |
This table doesn't crown a universal winner: it illustrates a trade-off. The scripted chatbot optimizes for safety and budget. The AI chatbot optimizes for experience and question coverage. The right choice depends on the nature of your business, the volume of requests, and your industry's tolerance for risk.
Which chatbot for which need? Three small and midsize business cases
Nothing beats concrete situations to help you decide. Here are three profiles we often come across, along with the solution that suits each one best.
The trades business and the local plumber
A plumbing company mostly gets requests for quotes and urgent repairs. The questions vary little: service area, availability, booking an appointment. A scripted chatbot that qualifies the request and then offers a time slot does the job perfectly, on a controlled budget. No need to roll out the AI artillery.
The online store with a large catalog
An online retailer selling hundreds of products gets unpredictable questions: compatibility between two products, materials, delivery times to a remote region, care advice. Scripting all of that is impossible. An AI chatbot connected to your product pages answers accurately, reduces cart abandonment, and saves the support team precious time. It's a textbook case where AI quickly pays for itself.
The consulting firm or services agency
A services business with a rich offering has prospects asking open-ended questions about methods, timelines, and ballpark pricing. An AI chatbot then acts as a smart front desk that pre-qualifies leads before handing off to a human. To dig deeper into that handoff, see how to automate customer service without dehumanizing the relationship.
The best chatbot isn't the smartest one; it's the one that matches exactly the question your customers are actually asking.
How much does each solution cost?
The budget question comes up fast, and the answer isn't a single figure but a cost structure. A scripted chatbot rests on a tool subscription and some setup time: most of the spend is up front, then the monthly cost stays low and stable. An AI chatbot adds two line items: design (connecting to your data, framing the answers, testing) and usage, billed according to the volume of exchanges with the model.
In other words, a lightly used AI chatbot costs little to run, but a heavily used bot generates a usage bill you need to anticipate. Conversely, a heavily used scripted chatbot costs no more, but grows more frustrating as requests diversify. To think in terms of return on investment rather than raw spending, the angle is well covered in our article on how an AI chatbot can cut your support costs.
Watch out for hidden costs
The real cost of a chatbot isn't limited to the subscription. Factor in the time to write the scripts or prepare the knowledge base, the maintenance when your offering changes, and the review of conversations to fix what's not working. A neglected chatbot quickly becomes counterproductive.
How to choose: a step-by-step method
Rather than starting from the technology, start from the need. Here's a simple approach to follow before you decide.
- 1List the real questions your team receives: emails, calls, messages. That's your raw material.
- 2Sort them: the ones that come back identically (scriptable) and the ones that vary endlessly (AI's turf).
- 3Estimate the volume: how many requests per month, at what times, with what level of urgency.
- 4Set the acceptable level of risk: is a rough answer inconvenient, or even dangerous in your industry?
- 5Define the escalation point: at what moment the conversation must switch to a human.
- 6Choose the technology accordingly: scripted if everything is well-defined, AI if variety dominates, hybrid when in doubt.
- 7Plan a testing phase on a limited scope before rolling it out across the whole site.
Start small
Don't try to cover every question from day one. Target the five most frequent requests, measure satisfaction, then expand. A chatbot that does a little well beats a chatbot that does a lot badly.
Common mistakes to avoid
- Choosing AI for show when three scripts would do: you're paying for needless complexity.
- Deploying an AI chatbot without guardrails, at the risk of made-up answers. The topic of generative AI hallucinations is worth understanding before you dive in.
- Hiding access to a human: a stuck customer who finds no way out leaves the site more annoyed than before.
- Forgetting maintenance: a bot showing outdated hours or prices damages your image.
- Neglecting tone: a chatbot should speak like your brand, not like an administrative form.
The best of both worlds: the hybrid approach
In practice, the most robust solution often combines both logics. You keep scripted buttons for sensitive, predictable actions (booking an appointment, tracking an order), and you switch on AI for everything else, with strict guardrails: answers limited to your knowledge base, automatic escalation to an advisor on delicate topics, and a deliberate refusal to answer outside its scope. If you're starting from scratch, our guide on building a chatbot for your website, step by step lays out the way forward.
The golden rule
A good chatbot knows its limits. It answers accurately within its scope, clearly says when it doesn't know, and hands off to a human rather than improvising. It's this programmed humility that builds trust.
Guide your customers to your chatbot
A QR code on your flyers, storefronts, or invoices sends your customers straight to your contact page or your online assistant. Generate yours for free in seconds.
Frequently asked questions
How do I know whether I need an AI or a rule-based chatbot?
Look at the questions your team actually receives. If they come back identically and can be counted on one hand, a scripted chatbot is enough. If customers phrase varied, unpredictable requests, an AI chatbot connected to your data will be far more useful.
How much does a chatbot cost for a small or midsize business website?
A scripted chatbot rests mainly on a moderate monthly subscription and limited setup time. An AI chatbot adds a design cost and a usage cost that varies with the volume of exchanges. The budget therefore depends on the technology chosen and how much traffic your site gets.
Why can an AI chatbot give wrong answers?
A language model can invent a plausible but false answer when it lacks the information. That's why you need to plug it into your knowledge base and set guardrails: limit it to your data, allow it to say "I don't know," and provide escalation to a human on sensitive topics.
Which chatbot needs the least maintenance?
None is fully autonomous. The scripted chatbot requires manually adding branches for every new case. The AI chatbot requires keeping its knowledge base up to date. In both cases, plan for regular review of conversations to fix what isn't working.
Will a chatbot replace my customer service?
No, it assists it. A well-designed chatbot handles repetitive questions and saves your team several hours a week, freeing them to focus on complex, high-value cases. The goal is to support humans, not to remove them.
In summary
The scripted chatbot shines through its reliability and controlled cost for well-defined needs; the AI chatbot excels as soon as questions diversify and the customer experience becomes a priority. Most often, it's a combination of the two, with clear guardrails and escalation to a human, that delivers the best result. The real issue isn't the technology, but the fit with the questions your customers actually ask. If you're still unsure which solution suits your business, let's talk about your project: TC Automation helps you design an assistant that is useful, reliable, and true to your brand.
