Chatbot Cost: CRM, RAG and Handoff 2026 | KUMO

Chatbot cost in 2026 is not decided by chat bubbles or model choice alone. The real price comes from CRM and helpdesk handoff, RAG over company documents, compliance controls, human escalation, usage volume, monitoring, and post-launch maintenance.

If you need a chatbot that can answer from real business data and hand off safely to sales or support, Book a 30-Min AI Scoping Call with KumoHQ before choosing a vendor or platform.

CRM handoff and RAG scope decide the real cost

A low-cost FAQ bot can answer static questions. A production chatbot needs customer context, permission-safe retrieval, escalation rules, analytics, fallback paths, and a clean handoff into CRM, helpdesk, WhatsApp, email, or internal operations systems.

• List the systems the chatbot must read from: website content, help docs, product catalogues, CRM records, tickets, invoices, or policy files.
• Decide which answers require human approval, especially pricing, refunds, medical, financial, legal, account, or support-risk responses.
• Budget for evaluation cases, monitoring, drift checks, support-team training, and monthly improvement instead of treating launch as the finish line.

Related planning guide: AI customer support automation RFP template

Related governance guide: AI governance checklist for CRM and ERP workflows

Related automation guide: Workflow automation requirements checklist

A chatbot can cost $8,000. It can also cost $87,000. Both quotes can be honest.

The price gap usually comes from six things: the data it needs, the tools it must connect to, the number of conversations it will handle, the compliance burden, the handoff design, and who maintains it after launch.

This guide is for owners and operators who need a real budget before they speak to vendors. Not a demo budget. A production budget.

KUMO has built production software for 13 years, including the core product behind Volopay, a YC S20 company that raised $31M. The lesson is simple: chatbot cost is rarely about the chat screen. It is about the system behind it.

Talk to a KUMO engineer: Cost depends on six variables. Send us your scenario and we will estimate the real range. Book a 30-Min AI Scoping Call

The short answer

Most serious chatbot projects fall into four bands:

If your total need is under $15,000, start with a SaaS tool. If the chatbot needs to read your CRM, create tickets, handle private data, and route work to humans, you are no longer buying a chatbot. You are building a workflow product.

That is where KUMO usually fits.

Cost driver 1: what the bot needs to know

A chatbot that answers from ten public web pages is cheap.

A chatbot that answers from your CRM, support history, invoices, product database, SOPs, and policy documents is a different build.

Typical ranges:

• Public FAQ knowledge base: $2,000 to $8,000
• Private document search: $8,000 to $25,000
• CRM and ticket history search: $15,000 to $60,000
• Multi-system answer engine: $50,000 to $150,000+

The hidden cost is not uploading documents. The hidden cost is answer quality. Someone has to decide what the assistant is allowed to answer, when it must say it does not know, and when it must hand the conversation to a human.

A bad chatbot answers fast. A useful chatbot knows when to stop.

Cost driver 2: integrations

Every integration adds cost because every business system has rules.

A simple integration might read a record from HubSpot or create a ticket in Zendesk. A complex integration might check customer status in Salesforce, pull billing data from Stripe, create a Jira task, and notify a Slack channel.

Typical ranges:

• Simple read-only integration: $2,000 to $5,000
• Write-back integration: $5,000 to $15,000
• Complex CRM or ERP integration: $15,000 to $50,000
• Multi-system workflow: $40,000 to $120,000+

This is why cheap chatbot quotes break later. The vendor prices the chat window. The business actually needs CRM, support, billing, permissions, logs, and error handling.

If your chatbot needs CRM or helpdesk handoff, Book a 30-Min AI Scoping Call before you compare vendor quotes.

Cost driver 3: conversation volume

Low traffic is forgiving. High traffic is expensive because model calls, monitoring, fallbacks, and latency all matter.

For under 10,000 conversations per month, model cost may be only a few hundred dollars. At 100,000+ conversations per month, the monthly bill can move into five figures depending on model choice, prompt length, retrieval, and testing.

Budget ranges:

• Under 10,000 conversations per month: $0 to $5,000 per month in operating cost
• 10,000 to 100,000 conversations per month: $2,000 to $25,000 per month
• 100,000+ conversations per month: $10,000 to $50,000+ per month

The right question is not, “Which model is cheapest?” The right question is, “Which conversations deserve an expensive model?”

A production system can route simple questions to cheaper paths and use a stronger model only when needed.

Cost driver 4: compliance and risk

Compliance changes the build.

A retail chatbot that recommends products does not carry the same risk as a healthcare chatbot that touches patient data or a finance chatbot that reads billing records.

Typical ranges:

• No regulated data: $0 to $10,000 extra
• GDPR consent, deletion, and audit needs: $10,000 to $40,000 extra
• HIPAA-style handling: $25,000 to $100,000+ extra
• SOC 2-aligned controls and audit logs: $30,000 to $150,000+ extra

Gartner predicted that over 40% of agentic AI projects will be cancelled by the end of 2027 because of cost, unclear value, or poor risk controls. That warning applies directly to chatbot projects. A demo that cannot pass legal review is not a project. It is a slide.

Cost driver 5: human handoff

Most chatbot failures happen at the handoff.

The bot gives an answer. The customer is not satisfied. Then what?

A real handoff needs:

• clear escalation rules
• a summary for the human agent
• ticket creation
• owner assignment
• service-level tracking
• audit history
• a way to correct bad answers later

Simple handoff can cost $5,000 to $15,000. A full support or sales handoff system can cost $25,000 to $80,000.

This is also where lead capture lives. If the bot identifies a qualified buyer but does not route them to sales, the traffic is wasted.

KUMO can map this handoff before build starts. Book a 30-Min AI Scoping Call if the chatbot has to create tasks, tickets, or sales follow-up.

Cost driver 6: maintenance

A chatbot is not done when it launches.

Products change. Policies change. Prices change. Teams change. Bad answers appear in logs. Users find edge cases.

Plan for 20% to 30% of build cost per year as maintenance. A $60,000 assistant usually needs $12,000 to $18,000 per year for fixes, model changes, retrieval tuning, monitoring, and small improvements.

If a vendor quotes no maintenance cost, ask who owns failed answers after launch.

Three budget examples

Small SaaS company

Need: qualify website visitors, answer pricing questions, route demo requests to the right person.

Likely scope:

• website chat
• CRM lookup
• lead scoring
• calendar routing
• basic analytics

Likely budget: $20,000 to $45,000.

This is a strong project if the team already has traffic and a clear sales process.

Mid-market retailer

Need: answer order questions, product questions, return policy questions, and route difficult cases to support.

Likely scope:

• ecommerce integration
• helpdesk integration
• returns workflow
• order-status lookup
• human handoff
• conversation analytics

Likely budget: $60,000 to $150,000.

The cost is not the chatbot. The cost is making the chatbot safe inside the support process.

Healthcare clinic group

Need: answer appointment, insurance, and admin questions without exposing private health data.

Likely scope:

• strict answer boundaries
• consent handling
• audit logs
• secure hosting
• human review
• compliance workflow

Likely budget: $120,000 to $350,000+.

This is not a good fit for a generic chatbot plugin.

When you should not hire KUMO

If your need is simple, do not overbuy.

Use a SaaS chatbot if:

• the bot only answers public FAQ questions
• no private customer data is needed
• there is no CRM write-back
• wrong answers are low risk
• the budget is under $15,000

Good tools can cover that range.

Hire a software partner when the chatbot has to become part of your business system: sales, support, billing, compliance, or operations.

What to do this week

Use this five-step worksheet before speaking to vendors:

1. List the top 20 conversations the chatbot must handle.
2. Mark which conversations need CRM, helpdesk, billing, or document access.
3. Mark which answers need human approval.
4. Estimate monthly conversation volume.
5. Decide who owns bad answers after launch.

If the worksheet shows private data, integrations, or human handoff, the chatbot is a software project. Book a 30-Min AI Scoping Call and KUMO will help size the first release.

Related KUMO guides

• AI integration services
• AI chatbot implementation checklist
• AI customer support automation RFP
• AI workflow automation ROI calculator
• AI implementation roadmap
• Volopay case study

CTA: get a real estimate

If you are comparing chatbot quotes, send us the workflow.

KUMO will help you separate the chat screen from the real system behind it. If SaaS is enough, we will say that. If custom build makes sense, we will map the budget range, timeline, risks, and first version.

Book a 30-Min AI Scoping Call

FAQ

How much does an AI chatbot cost in 2026?

A serious custom chatbot usually costs $15,000 to $120,000. Simple SaaS setups can be under $15,000. Regulated or high-volume assistants can pass $250,000.

Why do chatbot quotes vary so much?

Because vendors quote different things. One quote may cover a chat widget. Another may include CRM, support handoff, private data search, compliance, monitoring, and maintenance.

Can I build a chatbot with ChatGPT only?

You can build a demo. A production chatbot usually needs retrieval, permissions, logs, fallbacks, human handoff, and monitoring.

When should I use a SaaS chatbot instead of custom build?

Use SaaS when the bot only answers public FAQs and wrong answers carry low risk. Use custom build when the chatbot touches customer data or business workflows.

What is the first step?

Map the top 20 conversations the chatbot must handle. Then mark which ones need private data, integrations, compliance, or a human handoff.