Why Chatbot Pricing Is So Hard to Find Online
If you search "how much does a chatbot cost," you will find answers ranging from $500 to $500,000. Both numbers are technically accurate for different things. The $500 is a no-code bot with a free-tier backend and one integration. The $500,000 is an enterprise-grade conversational AI with custom model training, integration into five legacy systems, compliance review, and a year of managed support.
Neither of those is probably what you need. This guide breaks down what actually determines chatbot development cost, gives you realistic ranges for different types of projects, and helps you understand what you are paying for.
The confusion in pricing is also partly because "chatbot" covers at least three genuinely different things: a rule-based decision tree that routes users to the right page, an LLM-powered assistant that answers natural-language questions grounded in your data, and a fully autonomous AI agent that takes actions inside your business systems. Each costs a different amount to build and maintain. When agencies quote you, make sure you know which category they are actually quoting.
The Three Tiers of Chatbot Development
Tier 1: No-Code and Low-Code Tools ($0–$500/month, self-built)
Platforms like Intercom, Tidio, ManyChat, and Landbot let you build rule-based chatbots without writing code. These tools are best for simple FAQ handling, basic lead capture, or routing conversations to human agents.
The "cost" here is mostly your time and the platform subscription. The limitation is scope: these tools follow decision trees. They cannot understand nuanced natural language, cannot reason across complex queries, cannot access live data from your systems, and cannot handle anything that falls outside their predefined flows.
A concrete example: a dental practice sets up a Tidio bot to answer "What are your opening hours?" and "Do you accept insurance?" That works. The moment a patient types "I broke a filling last night, what should I do?" — the bot either returns an irrelevant canned response or escalates to a human. That escalation is fine if your front desk has capacity. If call volume is the problem you are trying to solve, it is not.
If your use case is simple and your team has time to manage a no-code tool, this tier is worth trying before investing in custom development.
Tier 2: LLM-Powered Chatbot with Standard Integrations ($8,000–$35,000, custom built)
This is the most common tier for businesses that have outgrown no-code tools and need a chatbot that understands natural language, can answer questions grounded in their own data, and integrates with one or two systems (a CRM, a booking calendar, a support ticketing platform).
A chatbot at this tier typically involves:
- An LLM backend (OpenAI, Anthropic, or similar) for understanding and response generation
- A retrieval-augmented generation (RAG) layer to ground responses in your company's knowledge base
- A web or messaging interface for users to interact with
- One to three integrations with business systems
- Conversation logging and basic analytics
The cost range reflects the complexity of the knowledge base, the number and depth of integrations, and how much conversation design and testing the project requires.
A real-world example at the lower end of this range: a 12-person law firm builds a chatbot to handle initial intake questions from prospective clients. It answers questions about practice areas, fee structures, and eligibility for a free consultation. It integrates with their Calendly to book discovery calls without any human involvement. Total project: around $11,000 including discovery, build, and one month of post-launch tuning. The firm gets back roughly 8 hours per week of front-desk time previously spent on email and phone triage.
At the higher end: an e-commerce company with 3,000 SKUs builds a chatbot that answers product questions, checks live inventory via their Shopify API, initiates returns through their support system, and escalates complex complaints to human agents with full conversation context. That project sits closer to $30,000 because of the integration complexity and the volume of testing required.
Tier 3: Custom AI Agent with Deep Integrations ($35,000–$120,000+)
A more complex deployment — a voice chatbot that handles inbound calls, a chatbot that manages multi-step workflows across multiple systems, or a chatbot that needs to handle high conversation volume with enterprise-grade reliability and compliance requirements.
The cost is driven by integration depth (the more systems it connects to, the more each one costs to build and test), voice infrastructure (telephony, STT, TTS pipelines), compliance requirements (healthcare, finance, legal), and the evaluation infrastructure needed to verify quality at scale.
A healthcare provider building a HIPAA-compliant appointment management voice bot that integrates with their EHR, pharmacy system, and billing platform is firmly in this tier. So is a financial services firm building a chatbot that helps advisors pull client portfolio data, run compliance checks, and generate meeting summaries — because every part of that workflow touches regulated data with specific security requirements.
What Drives Chatbot Development Cost
Knowledge base complexity
The chatbot needs to know things. If those things are in a well-structured, accessible format (a set of Notion pages, a clean FAQ doc, a product catalogue in a spreadsheet), building the knowledge layer is relatively fast. If your knowledge is scattered across PDFs, legacy systems, email threads, and people's heads, it takes significantly longer to structure, clean, and load.
A manufacturing company with 15 years of product documentation in mixed PDF formats, some scanned from physical manuals, will spend 30–40% of their project budget just getting that knowledge into a usable state before a single conversation is designed. That is not a development team being inefficient — it is the actual cost of making messy knowledge machine-readable.
Integration depth
Each integration with a business system adds cost: scoping what data flows in and out, building the API connection, handling authentication, dealing with error cases, and testing against real data. A chatbot that answers questions costs less than one that books appointments, checks inventory, updates records, and sends confirmation emails.
Integrations also compound in complexity. Connecting to one well-documented REST API is relatively simple. Connecting to a 15-year-old ERP system with no public API, requiring custom middleware and real-time sync, can cost as much as the chatbot itself.
Channel and interface
A chatbot embedded in a website is the cheapest to build. WhatsApp integration adds complexity (Business API setup, message template approval). Voice adds the most: telephony infrastructure, STT/TTS pipelines, latency optimisation, and interruption handling.
WhatsApp specifically has approval timelines that are outside any development team's control. Meta's Business API onboarding can take two to four weeks. If your timeline includes a hard launch date, build that into the project schedule.
Conversation design
How many distinct flows does the chatbot need to handle? How many edge cases? How much iteration is needed to get responses that are accurate, appropriately toned, and consistent? For a narrow-scope chatbot (appointment booking for a single service), this is fast. For a general customer support chatbot across a complex product line, it takes significantly longer.
A recruiter chatbot for an agency with 12 active job categories needs tested responses for each category plus edge cases: what happens when a candidate asks about a role that is already filled, wants to know salary before applying, or asks a question the chatbot genuinely cannot answer? Each of those paths needs a deliberate response.
Post-launch support
A chatbot is not set and forget. It needs monitoring, knowledge base updates as your products and policies change, and prompt tuning as you identify failure modes in production. Factor in the cost of ongoing maintenance when evaluating total cost of ownership.
Most teams budget $500–$2,000 per month for ongoing support depending on conversation volume and the frequency of knowledge base changes. A business that updates pricing quarterly can get by with light-touch support. A business that changes its service offerings frequently — a staffing agency, for instance — needs more active maintenance.
Off-the-Shelf vs Custom Built: A Practical Comparison
| Off-the-shelf (no-code) | Custom LLM chatbot | |
|---|---|---|
| Upfront cost | $0–$500/mo | $8,000–$120,000 |
| Setup time | Days | 4–12 weeks |
| Natural language understanding | Limited / keyword-based | Full LLM comprehension |
| Custom knowledge base | Basic FAQ uploads | Full RAG over your data |
| System integrations | Pre-built connectors only | Any API or database |
| Conversation quality | Template responses | Contextual, branded voice |
| Compliance support | None | Can be built in (HIPAA, GDPR) |
| Ongoing control | Platform-dependent | Fully owned |
| Best for | Simple FAQ, lead capture | Meaningful business workflows |
The honest answer is: if your use case is genuinely simple, start with a no-code tool. If you hit its limits within 90 days — which most growing businesses do — budget for a custom build from the start rather than paying twice.
What to Expect in Practice
Most chatbot projects follow roughly the same shape. Discovery takes one to two weeks: interviews with your team, mapping the most common queries, reviewing existing documentation, scoping the integrations. This is the phase most low-cost agencies skip. It is also the phase that determines whether the finished chatbot is actually useful.
Build takes four to eight weeks depending on scope. The first working version is rarely what ends up in production. Iteration on conversation quality — tone, accuracy, edge cases — is where most of the real work happens and where the difference between a $15,000 chatbot and a $25,000 one becomes visible.
Testing before launch should include real queries from your team and ideally from a small group of real users. The chatbot will fail on things you did not anticipate. That is expected. A good development team builds in time to find and fix those failures before they hit your customers.
The first 30 days after launch are when you learn the most. Conversation logs reveal what users actually ask, how the chatbot handles unexpected inputs, and where users drop off. Allocate time and budget for a post-launch tuning phase.
Common Mistakes That Inflate Cost (or Kill the Project)
Starting with the channel, not the use case. Deciding "we want a WhatsApp chatbot" before knowing what the chatbot needs to do leads to scope drift and wasted budget. Start with the problem, then choose the channel.
Underestimating knowledge base work. Teams regularly assume their existing documentation is ready to use. It almost never is. Expect to spend time cleaning, structuring, and filling gaps before the AI can reliably answer from it.
Skipping evaluation. A chatbot that has not been systematically tested against real user queries will embarrass you in production. "It looked good in the demo" is not a testing methodology.
Ignoring handoff design. When the chatbot cannot answer — and it will not always be able to — what happens next? A chatbot with no clear escalation path frustrates users more than no chatbot at all. Design the failure mode before you design the success mode.
Treating it as a one-time project. Businesses that do not budget for ongoing maintenance end up with chatbots that give outdated information, handle edge cases poorly as their business evolves, and quietly erode trust with customers.
What You Should Expect to Pay for Quality
A quality custom LLM chatbot from a skilled development team costs between $10,000 and $40,000 for a well-scoped initial build. Below $10,000, you are either getting a very narrow-scope chatbot, a team cutting corners, or both.
The warning signs for cheap chatbot development are: no discovery process before quoting, no discussion of your data and knowledge base, no mention of testing or evaluation methodology, and no clarity on what happens after handoff.
Quality chatbot development includes time spent understanding your use case before building, careful knowledge base structuring, systematic testing against real user queries, and a clear post-launch plan.
How to Scope Your Project to Get an Accurate Quote
Before you ask for quotes, define:
- What questions should the chatbot answer? (List the top 20 that drive the most volume)
- What data sources does it need access to? (Documents, databases, live APIs?)
- What should it be able to do beyond answering questions? (Book, update, route, escalate?)
- Where will it live? (Website, WhatsApp, phone, internal tool?)
- What does a successful deployment look like in 90 days?
A development team that receives clear answers to these questions will give you a much more accurate quote than one working from "we want a chatbot for customer support."
Also ask your shortlisted agencies: what is your testing methodology, what does post-launch support look like, and can I see examples of chatbots you have built in a similar industry? Agencies that cannot answer these questions clearly are likely to underdeliver.
Related guides
- Chatbot for small business: what works and what it costs
- AI chatbot developer: what your chatbot actually needs
- WhatsApp AI chatbot for business: build one that works
- From chatbot to AI agent: when it's time to upgrade
- AI agent development cost: pricing tiers
- Our AI chatbot development services
What We Build at Woyce
We build custom AI chatbots at Tier 2 and Tier 3. We do discovery before quoting, scope clearly, and include evaluation and post-launch support as standard.
Talk to us about your chatbot project — we will give you an honest quote based on what you actually need to build.
Frequently Asked Questions
How much does it cost to build a chatbot in 2026?
A custom LLM-powered chatbot built by a professional development team typically costs between $8,000 and $40,000 for the initial build, depending on scope. Simple FAQ bots on no-code platforms can cost $0–$500 per month in subscriptions. Enterprise deployments with voice, deep system integrations, or compliance requirements can reach $120,000 or more. The most important variable is what the chatbot actually needs to do.
How long does chatbot development take?
Most custom chatbot projects take four to ten weeks from discovery to launch. A narrow-scope chatbot with a clean knowledge base and one integration can be done in four to six weeks. A more complex build with multiple integrations, a messy knowledge base, and significant conversation design can take ten to twelve weeks. Rushed timelines are one of the most common reasons chatbots underperform at launch.
What is RAG and why does it affect chatbot cost?
RAG stands for retrieval-augmented generation. It is the technique that lets an LLM answer questions based on your specific documents and data, rather than just its training data. Building a RAG layer requires chunking and indexing your documents, building a retrieval system, and tuning how the model uses retrieved context. The complexity of your source materials — how many documents, how well-structured, how frequently updated — directly affects how long this takes and therefore what it costs.
Can I build a chatbot in-house instead of hiring an agency?
You can, if you have engineers with LLM integration experience on your team. The honest consideration is not whether it is technically possible but whether it is the best use of your team's time. Building and maintaining a production chatbot is ongoing work. Many businesses start by building in-house, hit quality or maintenance issues, and bring in specialists later — often spending more total than if they had outsourced from the start.
What ongoing costs should I budget for after launch?
Expect to budget $500–$2,000 per month for ongoing support depending on conversation volume and how often your business changes. That covers knowledge base updates, prompt tuning as you identify failure patterns, LLM API usage fees (typically $50–$400 per month for moderate volumes), and monitoring. If you are on a no-code platform, the subscription fee replaces most of this. For custom builds, these are real ongoing costs to plan for.
Will a chatbot replace my customer service team?
For most small and mid-size businesses, a chatbot handles the high-volume, low-complexity queries that currently occupy your team's time — appointment bookings, FAQ responses, order status checks, initial intake. That frees your team for the conversations that actually require human judgment. Full replacement is neither realistic nor usually desirable. The goal is that your best people spend their time on the work that benefits from having a person involved.
What industries see the strongest ROI from custom chatbots?
Industries with high inbound query volume and relatively predictable question patterns see the clearest returns: legal intake, healthcare scheduling, e-commerce customer support, real estate qualification, financial services FAQ, and recruitment screening. A 12-person law firm or a 50-person logistics company typically sees ROI in six to twelve months based on staff time saved. Businesses with very unpredictable or highly nuanced queries — bespoke consulting, for example — see less clear-cut returns.