It’s the first question every business owner asks. And it’s the one most vendors dodge.
How much does an AI chatbot actually cost?
The honest answer: anywhere from $0 to $200,000+, depending on what you’re building, how you’re building it, and — critically — whether you’ve done the groundwork that determines if any of it will actually work.
The internet is full of pricing guides written by chatbot vendors trying to sell you their platform. This isn’t one of those. We build custom AI agents for businesses, and we’ve seen what happens when companies overspend on the wrong approach and underspend on the things that actually matter.
Here’s the real breakdown — what things cost, what drives those costs, and where most businesses waste money without realizing it.
The Three Approaches (And What Each Actually Costs)
There are fundamentally three ways to get an AI chatbot into your business. Each comes with different price ranges, different tradeoffs, and different failure modes.
Option 1: Off-the-Shelf Platforms
Typical cost: $50 – $500/month
These are the plug-and-play tools — Intercom, Drift, Tidio, Zendesk AI, and dozens of others. You sign up, connect your knowledge base, configure some flows, and launch.
What you get:
- Pre-built chat widgets and interfaces
- Basic FAQ handling and canned responses
- Template-based conversation flows
- Integrations with common tools (CRM, helpdesk, email)
- Analytics dashboards
What you don’t get:
- Deep customization for your specific workflows
- Real-time data pulls from your internal systems
- Complex decision logic beyond simple if/then branching
- An agent that truly understands your business context
Who this works for: Businesses with straightforward, low-volume support needs. If your customers mostly ask the same five questions and you just need a faster way to answer them, a $100/month platform might be all you need.
Who this doesn’t work for: Any business where the chatbot needs to do things — look up orders, process returns, qualify leads based on custom criteria, or make decisions that require real-time data from multiple systems. Platforms are built for broad applicability, not for your specific process.
The hidden cost: Most businesses outgrow these within 6 to 12 months. The migration cost — rebuilding everything you’ve configured on a new platform or moving to a custom solution — is never factored into the initial budget. We’ve seen companies spend $5,000 to $15,000 on a platform over a year, only to scrap it all and start over.
Option 2: Low-Code / No-Code Builders
Typical cost: $200 – $2,000/month (platform fees) + $5,000 – $20,000 (setup and configuration)
This is the middle ground — tools like Voiceflow, Botpress, Stack AI, or Make.com paired with an LLM provider. They give you more flexibility than off-the-shelf platforms but don’t require building from scratch.
What you get:
- Visual flow builders for conversation design
- LLM integration (GPT, Claude, etc.) for natural language understanding
- API connectors to pull data from your systems
- More control over logic, branching, and escalation
- Faster time to launch than a fully custom build
What you don’t get:
- Full control over the underlying architecture
- Performance optimization for high-volume workloads
- Deep integration with complex or legacy systems
- Ownership of the codebase (you’re renting the platform)
Who this works for: Businesses that need more than a basic chatbot but have relatively standard workflows. If your process can be mapped as a flowchart without too many exceptions, a low-code builder can get you to 70-80% of what a custom solution would deliver at a fraction of the cost.
Who this doesn’t work for: Businesses with complex, multi-step processes that involve pulling from multiple internal systems, handling edge cases gracefully, or scaling to thousands of daily interactions. The platform’s flexibility ceiling becomes your performance ceiling.
The hidden cost: LLM API fees. Most low-code platforms charge you for the builder, but the actual AI processing — every message sent through GPT-4 or Claude — is billed separately. For a customer-facing chatbot handling 500+ conversations per day, API costs alone can run $500 to $3,000/month, and they scale with volume. Budget for this before you launch.
Option 3: Fully Custom AI Agent
Typical cost: $15,000 – $200,000+ (build) + $1,000 – $5,000/month (maintenance and hosting)
This is what we do at Replyant, and it’s what makes sense when the other two options can’t deliver the outcome you need. A custom AI agent is built specifically for your business — your workflows, your data, your systems, your edge cases.
What you get:
- An agent designed around your actual process (not a template) — built with the same rigor that goes into production agent design
- Deep integration with your existing tools and databases
- Custom decision logic for your specific use cases
- Full ownership of the solution
- Performance tuned for your volume and complexity
- Ongoing optimization based on real results
What you don’t get:
- Instant deployment (custom builds take 4 to 12 weeks)
- A cheap experiment (this is an investment, not a test)
Who this works for: Businesses where the AI agent is handling a core operational workflow — customer support, lead qualification, order management, appointment scheduling — and the outcome directly impacts revenue or cost. If getting this right means a 30-40% reduction in support costs or a measurable lift in conversion rates, the investment pays for itself.
Who this doesn’t work for: Businesses that haven’t done the process work yet. A $50,000 custom agent built on top of an undocumented, broken workflow will fail just as spectacularly as a $50/month chatbot — and it’s the primary reason 86% of AI agent pilots never make it to production. This is exactly why we start with process design and readiness before we write a single line of code.
The hidden cost: Underinvesting in the process design phase. We’ve seen companies spend $100,000+ on a custom build that delivers poor results because they rushed through discovery and went straight to development. The build is only as good as the blueprint — which brings us to the part most pricing guides leave out entirely.
The Cost Nobody Talks About: What Happens Before the Build
Here’s what every chatbot pricing guide on the internet gets wrong: they only price the technology.
But technology is only 20% of the equation. The other 80% — the process design, workflow mapping, and operational readiness — is where the actual value is created. And it has real costs that most businesses either skip or don’t budget for.
Process Mapping and Redesign: $3,000 – $15,000
Before you build anything, someone needs to document exactly how your workflow operates today. Every step, every decision point, every handoff, every exception. Then redesign it — eliminate unnecessary steps, clarify escalation rules, define what the agent handles vs. what stays with a human.
This is the work that separates a chatbot that handles 20% of inquiries from one that handles 65%+. It’s also the work that most vendors skip because it’s not as exciting as a product demo.
If you’re using an off-the-shelf platform or low-code builder, you’re probably doing this yourself (or not doing it at all — which is why most chatbots underperform). With a custom build, this should be included in the engagement. If a vendor quotes you a build without mentioning process design, that’s a red flag.
Data Cleanup and Integration: $2,000 – $20,000
Your AI agent is only as good as the data it works with. If your CRM has 30% duplicate contacts, your agent will give wrong answers 30% of the time. If your order data lives in three systems that don’t sync, the agent can’t give customers a straight answer about their shipment.
Data cleanup isn’t glamorous, but it’s often the difference between an agent that works and one that doesn’t. Budget for it. The range depends entirely on how messy your current state is — some companies need a weekend of cleanup, others need a month.
Training and Change Management: $1,000 – $5,000
Your team needs to know how the agent works, when to intervene, and how to escalate. Your customers need a transition that doesn’t feel jarring. Your processes need updated documentation that reflects the new AI-assisted workflow.
This is the cheapest line item on the list and the one most companies cut first. Don’t. The best implementations we’ve seen are the ones where the team actively participates in the design and is genuinely prepared for launch.
The Real Cost Comparison: A Worked Example
Let’s make this concrete. Take a 25-person e-commerce company handling 200 customer inquiries per day. Three support agents, averaging $45,000/year each. The goal: automate the repetitive 60% so the team can focus on complex issues.
Approach A: Off-the-Shelf Platform
| Line Item | Cost |
|---|---|
| Platform subscription (annual) | $3,600 |
| Setup and configuration (internal time) | $2,000 |
| Year 1 total | $5,600 |
Likely result: Handles 20-30% of inquiries. Customers get frustrated with rigid responses. Support team still overwhelmed. After 12 months, you’re back to evaluating options.
Approach B: Low-Code Builder + LLM
| Line Item | Cost |
|---|---|
| Platform fees (annual) | $6,000 |
| Setup and configuration | $12,000 |
| LLM API costs (annual) | $12,000 |
| Year 1 total | $30,000 |
Likely result: Handles 40-50% of inquiries. Better natural language understanding. Still limited by platform constraints on complex workflows. API costs grow with volume.
Approach C: Custom AI Agent (Process-First)
| Line Item | Cost |
|---|---|
| Process mapping and redesign | $8,000 |
| Custom agent development | $40,000 |
| Data cleanup and integration | $5,000 |
| Training and change management | $2,000 |
| Hosting and maintenance (annual) | $18,000 |
| Year 1 total | $73,000 |
Likely result: Handles 60-70% of inquiries with 90%+ satisfaction. Pulls real-time order data. Handles returns, tracking, and product questions with custom logic. Support team drops to two agents focused on high-value interactions.
The Math That Matters
That third option costs 13x more than the first. But look at the outcome:
- Support cost saved: One fewer full-time agent = $45,000/year
- Efficiency gain: Remaining agents handle 2x the complex volume
- Customer satisfaction: Faster resolution, accurate answers, 24/7 availability
Year 1 net cost after savings: $28,000 for Approach C vs. $5,600 for Approach A that doesn’t actually solve the problem.
By Year 2, the custom agent costs drop to maintenance only ($18,000/year) while saving $45,000+ annually. The ROI compounds. The off-the-shelf platform? You’re still paying the subscription and the full support team.
The cheapest option is rarely the most cost-effective. And the most expensive option is only worth it if the process work has been done first. (Want a more rigorous way to run these numbers for your own business? Our five-number ROI framework gives you a step-by-step calculation method.)
Five Questions to Ask Before You Spend Anything
Before you commit budget to any approach, answer these honestly:
1. Have we mapped the workflow we’re automating? If not, start there. No amount of technology will fix a process you haven’t documented. Our readiness checklist covers exactly what needs to be in place.
2. What does success look like in numbers? “Better customer service” isn’t a metric. “Resolve 60% of inquiries automatically with 95% accuracy within 90 days” is. Define this before you price anything — it determines which approach is actually worth the investment.
3. What systems does the agent need to connect to? If the answer is “just our FAQ page,” a platform might work. If it’s “our CRM, order management system, shipping API, and return processing tool,” you’re looking at custom work.
4. What’s our volume — and where is it going? 200 conversations a day is very different from 20. And if you’re growing, today’s volume is the wrong number to plan for. Price for where you’ll be in 12 months.
5. What’s the cost of not automating? This is the number most businesses forget to calculate. Three support agents spending 60% of their time on repetitive inquiries is $81,000/year in wasted capacity. That’s not a technology cost — it’s an operational cost you’re already paying. The hire-vs-automate framework helps you compare the real costs of both paths — including the hidden costs of hiring for work that a machine should be doing.
The Bottom Line
AI chatbot pricing isn’t really about the chatbot. It’s about what you’re building it on top of.
A $50/month tool built on a solid, well-designed process will outperform a $100,000 custom agent built on chaos. The technology is the easy part. The process design, data readiness, and operational preparation — that’s where the investment either pays off or doesn’t.
The companies that get the best ROI aren’t the ones that spend the most. They’re the ones that spend in the right order: process first, technology second, optimization ongoing.
If you’re trying to figure out what the right investment looks like for your specific situation, that’s a conversation worth having — before you commit to any platform, vendor, or build.