Custom chatbot development in the US in 2026 typically costs $8,000 to $75,000. Most production chatbots — customer support, lead qualification, or internal knowledge-base assistants — land at $20K-$50K.
The floor is a simple FAQ chatbot on a rules-based decision tree embedded in a website. The ceiling is a multi-channel AI chatbot with RAG over a custom knowledge base, CRM integration, and human-handoff logic. Most LLM-powered production chatbots land at $25K-$55K. The single biggest predictor of where a specific engagement lands is scope discipline, operators who lock the spec in the first two weeks save 20-40% of total project cost over the next three months. Operators who let scope expand mid-build pay the inverse penalty. Either way, the $8K to $75K range is descriptive, not prescriptive: it reflects what a competent US vendor charges in 2026 for the work as scoped, not what a finished engagement has to cost.
| Component | Low | High |
|---|---|---|
Discovery + conversation design + intent mapping | $2K | $10K |
LLM integration + prompt engineering | $3K | $15K |
Knowledge base / RAG pipeline Document ingestion, chunking, embeddings, and retrieval. Scale depends on corpus size. | $2K | $18K |
Widget + UI build (web embed or native) | $2K | $10K |
CRM + ticketing integration (Salesforce, HubSpot, Zendesk) | $2K | $10K |
Human handoff logic + live-agent routing | $1K | $8K |
Analytics + conversation quality monitoring | $1K | $6K |
Discovery + conversation design + intent mapping
LLM integration + prompt engineering
Knowledge base / RAG pipeline
Document ingestion, chunking, embeddings, and retrieval. Scale depends on corpus size.
Widget + UI build (web embed or native)
CRM + ticketing integration (Salesforce, HubSpot, Zendesk)
Human handoff logic + live-agent routing
Analytics + conversation quality monitoring
Website widget is the baseline. Adding SMS, WhatsApp, Slack, or Teams multiplies integration cost 1.3-2× per additional channel.
Chatbots retrieving from a 50-page FAQ are different from chatbots retrieving from 50,000 product documents. Ingestion pipelines and update cadence drive RAG infrastructure cost.
Production chatbots need graceful fallbacks, human handoff, and escalation triggers. This logic is 20-30% of total build cost but the most important part for user trust.
Anonymous chatbots are simpler. Authenticated chatbots that personalize responses based on customer data (order history, subscription tier) require backend integration and session management.
Healthcare and financial chatbots need audit logs, PII handling, and sometimes regulatory review. These requirements add 30-50% to the base build.
Inparlor chatbot projects start at $15,000. Most production LLM chatbots land at $30K-$60K with a knowledge pipeline and CRM integration included. We build for production reliability, not demo quality. The premium over the floor of the market reflects scope we don't itemize, measurement infrastructure, post-launch stability, and a documented handoff that survives whoever happens to be on our team six months from now. Our proposals are itemized line-by-line so you can see what you're paying for; we'd rather lose the deal on transparent pricing than win it by hiding the math.
Custom quote
itemized proposal within 48 hours
Chatbots, AI agents, and RAG assistants that ship to production, not demos.
Full AI Chatbots & Agents breakdownIntercom Fin, Zendesk AI, or Drift AI for $500-$2,500/mo SaaS. These work well for standard support use cases on top of your existing ticketing system. Custom builds win when you need proprietary knowledge retrieval, custom UI, or deep system integration. The honest framing: cheaper vendors exist at every tier, Fiverr at the bottom, offshore agencies in the middle, established US-based mid-market shops at the top. The cost-quality curve is real but rarely linear. Going from a $5K vendor to a $15K vendor usually produces a meaningfully different outcome; going from $15K to $45K often produces a refinement, not a transformation. Where you sit on that curve depends on the cost of being wrong, not the budget you have available.
(Build cost) ÷ (support ticket deflection × cost per ticket + lead qualification lift × value per qualified lead)
$35K chatbot deflecting 40% of 500 monthly support tickets. At $18 avg ticket-handling cost, that's $3,600/mo saved. Plus qualifying 80 additional leads/mo at 15% close rate × $5K ACV = $60K additional ARR. Payback in under 6 months.
We'll send back an itemized proposal, scope, line items, timeline, and the team that would actually run the engagement. No discovery call to schedule a discovery call.
See the AI Chatbots & Agents serviceChatbots, AI agents, and RAG assistants that ship to production, not demos. Scoped and quoted individually — itemized proposal within 48 hours.