How much does an AI chatbot cost in India in 2026?

Short answer: a focused AI chatbot that answers from your own documents and hands over to a human usually costs ₹3–6 lakh to build in India (roughly US$4k–8k), plus a monthly running cost that depends on how many conversations it handles. A simple FAQ widget can cost less; a multi-language bot wired into your CRM, payments and WhatsApp costs more.

This guide breaks that number down so you can sanity-check any quote you receive — including ours.

typical build, focused chatbot

₹3–6Ltypical build, focused chatbot

to go live on your site

2–4 wksto go live on your site

running cost that scales with usage

Monthlyrunning cost that scales with usage

What you are actually paying for#

A production chatbot is more than a prompt on top of ChatGPT. The build cost covers five pieces of work:

1. Knowledge setup — collecting your FAQs, policies, product data and PDFs, cleaning them and making them searchable, so answers come from your content and not the model’s imagination.
2. Conversation design — what the bot should and should not answer, its tone, fallbacks, and exactly when it hands over to a person.
3. Integrations — the website widget, WhatsApp, your CRM (Zoho, HubSpot, Salesforce), order or booking lookups, and lead capture into Google Sheets.
4. Admin dashboard — conversation logs, unanswered questions, cost per conversation, and a way for your team to update content without a developer.
5. Testing and guardrails — an evaluation set of real customer questions, checks for wrong or risky answers, and load testing before launch.

Typical price ranges#

The monthly running cost#

Once the bot is live, you pay for three things every month:

• AI model usage — charged per token by OpenAI, Anthropic or Google. For most business chatbots this is the smallest line on the bill at moderate volume.
• Hosting and database — servers, the vector database that stores your knowledge, and logs.
• WhatsApp messaging, if you use it — Meta charges per message for business-initiated templates, while replies inside the customer-service window cost less. Check Meta’s current rate card for your category.

Five things that move the price#

1. Where the answers come from

Ten clean FAQs are cheap. Two thousand pages of PDFs, spreadsheets and old emails take real work to clean, split and keep up to date — and that work decides whether the bot is accurate.

2. How many channels

A website widget is one build. Adding WhatsApp, Instagram or an in-app chat each adds integration, testing and edge cases.

3. Languages

Hindi, English and the Hinglish your customers actually type need separate testing. Bilingual bots cost a little more and are usually worth it in India.

4. Actions, not just answers

Answering “what are your timings?” is simple. Checking an order, rescheduling a booking or creating a ticket means secure connections to your systems — that is where most of the budget goes.

5. How wrong is too wrong

A café can live with an occasional odd reply. A clinic, lender or insurer cannot. Stricter guardrails, approvals and audit logs add cost and are non-negotiable in regulated work.

How to keep the cost down#

• Start with one workflow — the question your team answers fifty times a day — and prove it before adding more.
• Reuse what you have. Existing FAQs, policies and product sheets are the fastest training material.
• Launch on one channel, usually WhatsApp or the website, then expand.
• Use the right model for each step. Not every reply needs the most expensive model.
• Insist on a fixed price for a clearly written scope, paid by milestone.

Chatbot or something simpler?#

Sometimes a well-structured FAQ page, a form or a WhatsApp quick reply beats an AI chatbot. We will tell you when that is the case — it is one of the principles we work by.