How Much Does an AI Chatbot Cost in 2026?

The Short Answer

A production AI chatbot costs $3,000–$25,000+ to build, depending on complexity. Here's the real breakdown.

Tier 1: Simple FAQ Bot ($3,000–$5,000)

A bot that answers predefined questions from a knowledge base. Think customer support for a small business.

What you get:

Chat widget on your website

Answers from a static set of Q&As

Basic conversation flow

Fallback to human support

Tech: OpenAI API + simple prompt engineering + a web widget. No vector database needed.

Timeline: 1–2 weeks.

Tier 2: RAG-Powered Knowledge Bot ($8,000–$15,000)

This is where it gets interesting. The bot reads your actual documents - PDFs, Google Docs, Notion pages - and answers questions with source citations.

We built Knoah as exactly this. Upload your docs, ask questions, get answers with page-number citations.

What you get:

Document ingestion (PDF, DOCX, URLs)

Vector embeddings for semantic search

Natural-language answers with citations

User authentication and usage tracking

Admin dashboard

Tech: Claude/GPT API + vector database (Supabase pgvector or Pinecone) + document processing pipeline + Next.js frontend.

Timeline: 3–4 weeks.

Tier 3: Full Conversational AI Product ($15,000–$25,000+)

A product where AI is the core feature - not just a support widget. Think AI-powered onboarding, AI content generation, or AI analysis tools.

What you get:

Everything in Tier 2

Custom AI pipeline (multiple models, chained prompts)

Streaming responses

User-specific context and memory

Billing integration (Stripe)

Production infrastructure (rate limiting, error handling, monitoring)

Tech: Multiple AI APIs + custom orchestration + production-grade infrastructure.

Timeline: 4–8 weeks.

The Hidden Costs Most Agencies Don't Tell You

API costs - Claude/GPT API calls cost $0.003–$0.06 per query depending on model and context length. Budget $50–$500/month for a typical B2B product.

Vector database - Supabase pgvector is nearly free at small scale. Pinecone starts at $70/month.

Hosting - $5–$50/month on Railway or Vercel for most apps.

Ongoing maintenance - Models update, APIs change, edge cases surface. Budget $150–$300/month or handle it yourself.

What We'd Recommend

For most businesses, Tier 2 (RAG-powered) is the sweet spot. You get genuinely useful AI that understands your specific business, without over-engineering.

Want to talk specifics? Get a free quote and we'll scope your chatbot in 30 minutes.