Truly understanding how much does a chatbot cost means looking beyond the subscription price. In 2026, the true investment is measured in TCO—Total Cost of Ownership—that is, the entire project lifecycle, which includes the various phases of development, maintenance, integrations, security, and data governance.
This is where the hidden costs that many companies initially overlook emerge: the list price of a platform or a language model covers only part of the spend, and the rest shows up once the chatbot goes into production and starts handling real conversations every day.
The technologies available today—no-code platforms, development frameworks, and LLM models—offer different approaches, with very different impacts on costs, control, and management complexity. This is especially true when it comes to sensitive data: choosing an advanced platform simplifies governance, while creating a chatbot independently with an LLM, even a business version, increases liability, risk, and unpredictability of expenses.
In this article, we look at how to realistically calculate the cost of a chatbot, the elements that make up the TCO, what you actually spend with a pay-as-you-go model like Crafter.ai's, and the technology choices that keep a project sustainable over time.
Table of Contents
- What is TCO and how it determines chatbot cost
- Items determining how much does a chatbot cost
- How much does a chatbot cost based on the technology used
- How much does a chatbot cost with Crafter.ai: a pricing example
- How to really estimate how much does a chatbot cost
- Data security and protection
- Conclusion: How much does a chatbot cost in 2026?
- Frequently asked questions about chatbot cost
What is TCO and how it determines chatbot cost
Budget overruns are well documented. According to a Gartner report covered by Campus Technology, at least 50% of generative AI projects will overrun their budgeted costs through 2028, due to poor architectural choices and lack of operational know-how. The same report expects inference, the day-to-day use of the model to answer users, to account for at least 70% of a model's lifetime costs, and notes that a production-ready system can be orders of magnitude more expensive than a pilot. This is where the concept of TCO—Total Cost of Ownership—comes in.
TCO allows you to calculate how much a chatbot costs and how much it costs to maintain it over time, answering the question every manager should ask before creating a chatbot:
"How much will it cost me not just today, but every month, every year, for the entire duration of the project?"
To truly understand how much a chatbot costs, you need to consider all the components that impact the project, including:
- Technical infrastructure and governance: This includes servers, hosting, APIs, language models, and control tools. It determines the stability of the chatbot and defines who can access, modify, or monitor the system.
- Development and initial configuration: This includes designing flows, training the model, customizing the tone, and setting up basic functionality.
- Ongoing maintenance: This involves reviewing content, correcting errors, optimizing responses, and monitoring performance over time.
- Integration with business systems: This refers to connecting the chatbot to CRM, ERP, ticketing, e-commerce, or internal databases, which is essential for achieving real automation and personalized responses.
- Security and compliance: This includes data protection, encryption, permission management, and compliance with GDPR and internal policies to prevent legal risks and unauthorized access.
- Updates and improvements over time: This covers the evolution of the chatbot through new features, modules, languages, or optimizations based on user feedback.
- Data management: This includes collection, storage, retention, auditing, conversation tracking, and classification of sensitive information.
Items determining how much does a chatbot cost

Creating and managing a chatbot involves a series of costs that go well beyond initial development and the platform subscription. Determining how much does a chatbot cost depends largely on the chosen technology: from initial configuration to conversational flow design, from model training to integration with databases, CRMs, or other business systems.
Each solution—no-code platform, custom framework, or LLM—requires specific activities that directly impact the TCO. Maintenance costs are also added, as a chatbot exists in a constantly evolving company: new services, internal policies, commercial offerings, and processes require constant updates to maintain the accuracy and quality of responses.
Recurring operating costs, related to licensing, API usage, conversation monitoring, data storage, and reporting to continuously optimize the conversational flow, should not be overlooked.
Internal team training is also essential: monitoring, updating, and interpreting the data generated by the chatbot helps improve its performance and make it an effective tool for the company.
Finally, the speed of technological change weighs on the budget: new models come out every few months, API prices change, and AI agents run multi-step workflows with several model calls per request. Keeping up takes time, resources, and expertise, and that is part of the project's true cost. In short, estimating how much a chatbot costs means adding up initial, recurring, and strategic expenses, all of which you need for a sustainable return on investment. For a step-by-step method to measure that return, see our guide on how to calculate the ROI of a chatbot.
How much does a chatbot cost based on the technology used

Choosing the right technology can significantly simplify TCO and transform hidden costs into predictable expenses. No-code platforms, for example, allow you to reduce development costs, speed up releases, manage updates, and integrate AI models in a secure and governed manner. At the same time, using LLMs or customized frameworks can also be effective, but requires careful planning and oversight to avoid inflating the budget. Understanding the differences between available technologies is the first step to maximizing your chatbot's ROI and keeping the project sustainable over time.
Development Frameworks
Creating a chatbot through custom development offers maximum freedom and customization. However, it comes with significant challenges: it requires a dedicated team, development times are long, initial costs are high, and maintenance increases exponentially with the project's complexity. This approach only makes sense if you require highly specific functionality or solutions unavailable elsewhere.
No-code Platforms
No-code platforms are currently the most competitive solution for creating a chatbot with controlled costs and a predictable ROI. They allow you to:
- reduce development costs, without the need for expert programmers;
- rapidly release the chatbot to the market;
- autonomously update workflows and content, without complex IT interventions;
- integrate AI models in a governed and secure manner.
Thanks to templates, automation, and integrated governance, no-code platforms give you the best balance between TCO and generated value, and let companies scale projects without costs getting out of hand. To dig deeper, read how no-code conversational AI works.
LLM (Large Language Models)
Integrating models like GPT allows for powerful, flexible, and conversational chatbots. At first glance, it may seem like the most cost-effective route: it doesn't require complex development and allows for rapid releases. It offers flexible responses and powerful natural language understanding.
However, the hidden costs are significant: token consumption can become unpredictable, continuous model supervision, filters, and moderation controls are required, as well as variable costs for APIs and internal resources dedicated to monitoring.
The business and enterprise tiers of the main LLMs improve the situation: data protection, compliance certifications, and admin dashboards help manage security and governance. But two problems remain:
- Spending remains variable and difficult to predict;
- Vendor lock-in is strong: the company depends entirely on a single model provider.
For this reason, relying exclusively on an LLM without a platform that controls governance, security, and costs remains risky.
Why token cost is the most underestimated risk
The per-token price you see on a price list is low, and that is exactly why it misleads. In an interview with CX Today, a Gartner analyst explains that LLM vendors currently keep prices low to grow their user base, and that those prices will have to change once these companies pivot to profitability. The same article notes that frontier models consume three, five, or even ten times more tokens for similar interactions. Gartner predicts that in customer service, the GenAI cost per resolution will exceed $3 by 2030, higher than many B2C offshore human agents.
For anyone estimating how much a chatbot costs, the takeaway is practical: a budget based only on today's token price may already be wrong next year. Choose a pricing model where you set the maximum spend, measure how many tokens a typical conversation really uses, and use the largest model only where it's needed, leaving simple requests to lighter models.
How much does a chatbot cost with Crafter.ai: a pricing example
Many articles on chatbot cost stop at spending categories without giving a number. Here is the number, because Crafter.ai's pricing is public and pay-as-you-go, with no subscription.
You start with a free trial of 1 million tokens, no credit card required. When the chatbot is ready for real traffic, you move to the €20 top-up, worth 100 million tokens: according to the Crafter.ai pricing page that is enough for thousands of conversations with your AI Agents. The credit never expires, so a quiet month doesn't mean money lost. You can top up manually whenever you need, or turn on auto top-up, which adds another €20 when your credit runs out and can be switched off at any time from your account.
Tokens cover conversations with agents, LLM calls, knowledge base (RAG) queries, and the platform's advanced features. For a project, this means variable spending has a ceiling you set: if you don't enable auto top-up, you never spend more than you loaded.
For an estimate based on your own volumes, use the AI agent traffic estimator: enter your expected conversations and your current service cost to get the estimated monthly spend and the breakeven point. The TCO point still holds: on top of tokens, count the team's time to configure and update the agent and any integrations with your business systems.
How to really estimate how much does a chatbot cost
Here's a practical 5-step model to accurately estimate your chatbot's TCO:
1. Define the objectives and context
First, clarify what the chatbot will be used for: customer support, sales, marketing, HR, or IT. The complexity of the project depends on the number of channels, languages, and features required.
2. Estimate initial development costs
Consider using a no-code platform, a development framework, or a Large Language Model (LLM). Consider licensing, development hours, and customizations required for your use case.
3. Estimate maintenance and upgrades
This includes periodically reviewing conversation flows, adding new content and features, monitoring the quality of responses, and adapting to changes in business processes.
4. Consider integration and infrastructure
An effective chatbot must communicate with CRM, ERP, e-commerce platforms, or ticketing systems. This includes technical integration costs, hosting, storage, and the APIs of LLM models. Also estimate token consumption per conversation: multiplied by the expected monthly volume, it is the item most often missing from the first quote.
5. Evaluate security, governance, and compliance
Data protection, permission management, GDPR, and local regulations are essential. Choosing a platform with integrated security tools reduces risks, while standalone solutions require additional resources and costs.
Data security and protection
Creating a chatbot means managing sensitive information, from customer data to conversation history, from sensitive logs to internal company information. The difference between using a dedicated platform and developing the project "by hand" with an LLM, even a business version, has a big impact on the final cost.
If you choose an advanced platform, data governance is already integrated: security, privacy, tracking, and auditing are automatically managed, reducing risks and simplifying the team's work. Conversely, relying directly on an LLM means that all responsibility falls on the company: privacy, storage, retention, role management, auditing, and conversation monitoring must be designed and controlled internally.
In practice, developing a chatbot independently with ChatGPT involves greater complexity, requires a dedicated team, and significantly increases the TCO, making it more difficult to predict the costs and risks associated with data management.
Conclusion: How much does a chatbot cost in 2026?
Answering the question "How much does a chatbot cost?" means evaluating the entire project lifecycle.
The expense of a chatbot doesn't end with the initial implementation: it's an ongoing investment. Creating a chatbot is a strategic choice that, when supported by the right tools, can reduce operating costs, speed up response times, and ease the workload of company teams.
However, the TCO (Total Cost of Ownership) depends largely on the technology chosen. No-code platforms have predictable costs, integrated governance, and immediate scalability. Development frameworks allow for maximum customization, but come at the expense of higher costs and longer lead times. LLMs like GPT are powerful and flexible, but they involve variable costs and more complex data management, especially without a governance platform. And with token prices likely to change, having a spending cap set in advance matters more than today's list price.
With a structured and informed approach, creating a chatbot stops being a gamble and becomes a sustainable and measurable investment. To test it on your own use case, start with the free trial and move to the €20 top-up only when your agent is ready.
Frequently asked questions about chatbot cost
How much does a chatbot cost for a small business?
With a pay-as-you-go no-code platform you can start for free and then spend €20 for 100 million tokens, with no subscription. On Crafter.ai the credit never expires, so a small business pays only for the conversations the chatbot actually handles. Add the internal time needed to prepare content and configure the agent.
Is an AI chatbot cheaper than a human agent?
It depends on the type of request. For repetitive first-level questions, usually yes. For complex cases, not always: Gartner predicts the GenAI cost per resolution in customer service will exceed $3 by 2030, higher than many offshore agents. That's why it makes sense to automate simple requests and hand complex ones to a person.
How much does TCO impact the real cost of a chatbot?
Total Cost of Ownership (TCO) includes the platform subscription or license plus development, maintenance, integration with enterprise systems, data security, updates, and team training. Ignoring these hidden costs inflates the budget: according to Gartner, at least half of generative AI projects will overrun their budget through 2028.
What are the most common hidden costs of running a chatbot?
Beyond initial development, hidden costs include: maintenance and updating of conversational flows; integration with CRM, ERP, or ticketing systems; team training and management; response quality monitoring and ongoing optimization; security and regulatory compliance for sensitive data; token consumption that grows with traffic. Considering these elements is essential to correctly estimate the cost of a chatbot and ensure a sustainable return on investment.
How much does the choice of technology affect the cost of a chatbot?
The technology chosen to create a chatbot has a direct impact on TCO. No-code platforms reduce initial time and costs, while development frameworks require dedicated technical teams and larger budgets for complex customizations. LLMs offer flexibility and advanced responses, but incur variable costs related to token consumption and response supervision. Furthermore, each choice affects ease of integration, maintenance, and data security, all of which impact the project's actual cost.
Is there a transparent pricing model for enterprise chatbots?
Yes. Crafter.ai uses an on-demand token model with public pricing: a free trial with 1 million tokens and a €20 top-up for 100 million tokens, credit that never expires, and optional auto top-up. This lets you plan your budget in advance and grow spending with traffic, without the surprises of uncapped consumption-based models.





