AI Chatbot Cost in India 2027: Build Quotes vs Monthly Bills

The AI chatbot cost in India has two parts that get mixed up: a one-time build quote and a monthly bill. As of 1 Nov 2026, running a WhatsApp bot for a clinic answering 300 chats a month comes to about ₹859 to ₹1,352 a month with GST, on the illustrative assumptions below. That covers Meta's fees, AI model usage and one small server. It excludes build, upkeep, staff time, card forex charges and any platform fee.

Your front desk answers the same questions all day: what time does the doctor sit, what's the fee, is Saturday open? So you asked around, and now you hold quotes that differ by lakhs.

This post gives you the sum to redo with your own numbers, what to ask each seller, and when not to build at all. The clinic is an example. The arithmetic works for any business answering a few hundred chats a month, and if you're still deciding what WhatsApp automation could do at all, start there.

One note on the "2027" in the title. No 2027 price list exists. Every price here is as of 1 Nov 2026, and we predict nothing.

AI chatbot cost in India: the monthly sum at 300, 1,000 and 3,000 chats

These are illustrative assumptions, not a real clinic. Measure your own in a pilot:

• Each chat gets 6 bot replies, so replies = chats × 6.
• Each reply reads 3,000 input tokens and writes 150 output tokens. Tokens are the units AI models bill by.
• All replies go from one number directly on Meta's Cloud API, with no platform in between.
• USD converts at ₹95.2492, the FBIL reference rate for 30 Oct 2026.

Line (monthly, as of 1 Nov 2026)	Formula	300 chats	1,000 chats	3,000 chats
Bot replies	chats × 6	1,800	6,000	18,000
Input / output tokens	replies × 3,000 / × 150	5.4M / 0.27M	18M / 0.9M	54M / 2.7M
Billable WhatsApp replies	max(replies − 1,000, 0)	800	5,000	17,000
WhatsApp, before GST	billable × ₹0.115	₹92.00	₹575	₹1,955
WhatsApp, with GST	× 1.18	₹108.56	₹678.50	₹2,306.90
Claude Haiku 5.5, before GST	see token table	₹64.29	₹214.31	₹642.93
One $6 server, before GST	$6 × ₹95.2492	₹571.50	₹571.50	₹571.50
Total with Haiku 5.5	(WhatsApp + model + server) × 1.18	₹858.79	₹1,605.75	₹3,739.92
Total with Gemini 3.5 Flash-Lite	same	₹1,040.86	₹2,212.67	₹5,560.70
Total with Gemini 3.8 Flash, current rate	same	₹1,351.91	₹3,249.51	₹8,671.21

Totals are computed before rounding, so redoing them from the rounded lines can differ by a paisa. Keeping one server at every volume is an assumption, and it is least safe at 3,000 chats. The token model leaves out caching discounts, retries and billed thinking tokens.

What would change these numbers

• Your replies per chat and prompt size. These two assumptions drive the token line.
• Staff replying on the same number, which can use up the free 1,000 (next section).
• A new Meta rate card, the rupee, or a model price. Check the dated pages cited here before you budget.
• Adding a platform, which can outweigh everything else in this table.

How Meta's WhatsApp bill works

Meta's India INR rate card effective 1 Oct 2026, per its pricing docs (we read the rate card PDF on 30 Oct 2026), before GST:

Category	Per delivered message
Marketing	₹0.8631
Utility	₹0.1150
Authentication	₹0.1150
Authentication, international	₹2.4971
Service	₹0.1150

What Meta states:

• The first 1,000 service messages per business phone number per month are free, with no rollover. Charges start at the 1,001st.
• Patient messages to you are free. Undelivered messages aren't charged.
• Utility templates sent inside an open window are charged and don't count in the free 1,000.
• A free entry point window runs "up to 7 days", tied to click-to-WhatsApp ads. WhatsApp's marketing page still says service is free with a 72-hour window. It conflicts with the dated docs, so go by the docs.
• Eligible India businesses must move to INR billing by 31 Dec 2026, with non-INR messages stopped from 1 Jan 2027.

Our reading, not Meta's words:

• Meta's definition of service messages includes replies written by a person, so staff replies on the same number use the same 1,000. If staff use all of it, 1,800 bot replies cost ₹207, or ₹244.26 with GST.
• If your code splits one answer into two sends, assume two charges.

Other lines to add: one utility reminder per chat adds 300 × ₹0.115 = ₹34.50, or ₹40.71 with GST. Our guide to appointment reminder templates covers those, and this breakdown of WhatsApp Business API message fees covers the rest.

Meta Business Agent is priced differently. Per Meta's docs, this separate category costs $2 per 1M tokens at about 20,000–25,000 tokens per message (4–5 cents), with no free tier. So 1,800 messages cost $72–$90, or ₹6,858–₹8,572 at the reference rate before tax. The conversion is illustrative.

GST: Meta's docs don't mention tax. Free Press Journal says the rates exclude taxes and cites 18%. Secondary tax sources describe clinical-establishment healthcare as exempt (entry 74, Notification 12/2017), and we haven't established whether input credit is available. Don't assume this GST is recoverable. Ask your CA.

LLM token cost by model

Formula: input tokens ÷ 1M × input price, plus output tokens ÷ 1M × output price. Monthly, before GST, from published prices as of 1 Nov 2026:

Model	Price per 1M, input / output	300 chats	1,000 chats	3,000 chats
Claude Haiku 5.5 (prompts up to 100K tokens)	$0.10 / $0.50	$0.675, ₹64.29	$2.25, ₹214.31	$6.75, ₹642.93
OpenAI gpt-6-luna	$0.10 / $0.50	$0.675, ₹64.29	$2.25, ₹214.31	$6.75, ₹642.93
Gemini 3.5 Flash-Lite	$0.30 / $2.50	$2.295, ₹218.60	$7.65, ₹728.66	$22.95, ₹2,185.97
Gemini 3.8 Flash, through 31 Dec 2026	$0.75 / $3.75	$5.0625, ₹482.20	$16.875, ₹1,607.33	$50.625, ₹4,821.99
Gemini 3.8 Flash, published rate from 1 Jan 2027	$1.50 / $7.50	$10.125, ₹964.40	$33.75, ₹3,214.66	$101.25, ₹9,643.98
OpenAI gpt-5.4-mini	$0.75 / $4.50	$5.265, ₹501.49	$17.55, ₹1,671.62	$52.65, ₹5,014.87

The January Gemini row is Google's own published rate, shown at today's exchange rate for comparison only. Google recommends 3.5 Flash-Lite or 3.8 Flash for new projects. Its 2.5 models are available only to those who used them before, so don't plan on them. We left out OpenAI's older mini and nano models because they have shutdown dates.

Hosting cost for a WhatsApp chatbot

The bot needs somewhere to receive messages and run its code. List prices as of 1 Nov 2026, converted at the same rate, before GST:

• DigitalOcean 1 GiB droplet: $6 = ₹571.50, or ₹674.36 with GST. This is the server in our table.
• DigitalOcean $4 droplet: ₹381.00.
• AWS Lightsail $5: ₹476.25. Mumbai plans get half the transfer allowance.
• Cloudflare Workers: free for 100,000 requests a day, or $5 paid (₹476.25).

At 300 chats, the server is the largest line in our table when you use Haiku 5.5.

Platform (BSP) fees: the line that can outweigh the rest

A BSP (business solution provider) is a company reselling WhatsApp access with an inbox and bot tools on top. Sellers' prices, checked 1 Nov 2026, excluding Meta charges unless stated. AI message and credit definitions aren't comparable across them.

Seller	Plans, monthly	AI allowance	Service rate
Interakt	Growth ₹2,799 + taxes, Advanced ₹3,799	AI ₹2,499 with 100 AI messages, then ₹0.50 each; also lists a ₹3,000 add-on	₹0.128 / ₹0.127
Wati	Growth ₹2,699 billed quarterly, Pro ₹5,799, Business ₹16,799	Growth has 2,000 AI credits; credits per reply not established	Not in our check
Gallabox	Basic ₹2,999 quarterly or ₹2,399 annual, Essential ₹6,999, Advanced ₹16,999	Basic has 500 AI credits	₹0.1323
AiSensy	Plan cards didn't load for us, so omitted	AI Agent Builder ₹1,350 with 1,000 AI messages; overage not stated	₹0.145

Gallabox's FAQ says it applies a markup, and doesn't size it. To compare, here's an illustration: if Interakt's Growth and ₹2,499 AI plans stack and one reply equals one AI message, 1,800 replies come to about ₹6,250 before taxes (₹2,799 + ₹2,499 + 1,700 × ₹0.50 + 800 × ₹0.128). Interakt's page doesn't say how its AI options combine, so ask them to confirm.

A BSP can still be the right choice, since you get an inbox and a bot without commissioning code. Compare a year of plan fees against a build.

Build quotes: why the AI chatbot cost in India differs by lakhs

These are sellers' own estimates, checked 1 Nov 2026, and their tiers aren't defined alike:

Seller	Published estimate
Beeps Digital	Custom WhatsApp + n8n + GPT: ₹25,000–₹60,000
Scallar	Basic FAQ ₹30,000–₹75,000, business ₹1–4 lakh, advanced ₹5 lakh+
Dynamb	Rule-based ₹30,000–₹75,000, ₹1–4 lakh, RAG ₹5–25 lakh
Codingclave	Rule-based ₹35,000–₹1.25 lakh, AI ₹2.5–6 lakh, WhatsApp AI ₹2–5.5 lakh
WOWinfotech	₹50,000–₹80,000, ₹5–6 lakh+, ₹7–10 lakh+ by tier
Cyfuture AI	Custom: ₹10 lakh–₹1 crore+

So the range runs from ₹25,000 to over ₹1 crore under one label. An FAQ bot and an agent that reads orders and enters them in your system are different projects.

Sellers' monthly figures of ₹5,000–₹50,000 sit far above the direct costs in our table. They may include platform fees, monitoring, upkeep, margin or higher volumes. Ask what each figure covers.

Questions to ask each seller:

• Channels: WhatsApp only, or website too? Does it answer missed calls?
• Integrations: does it read your appointment book, or only a document?
• Handoff: how does a chat reach a person, and what does the patient see?
• Knowledge upkeep: who changes timings and fees, how fast, and at what cost?
• Accounts: who holds the WhatsApp and LLM accounts, and whose card pays?
• Code: who owns it if you leave?

What a clinic chatbot must not do

We aren't lawyers and haven't verified all of these against the originals. Take this list to your own lawyer.

• WhatsApp's Business Messaging Policy (last updated 23 Sep 2026) says not to use WhatsApp for telemedicine, or to send or request health information "if applicable regulations prohibit distribution" to systems lacking heightened safeguards. That is conditional, and whether it applies to your clinic is a legal question. Automation also needs prompt, clear escalation paths to a person.
• Meta's platform terms, s.4.7, bar AI providers where AI is the primary function, and Meta decides; TechCrunch reported this as effective 15 Jan 2026. Our reading: a clinic's own bot about clinic services appears to fall on the incidental side. Meta's pricing docs price third-party AI replies as service messages.
• Telemedicine Practice Guidelines (25 Mar 2020). We read summaries, not the original. They say AI can't counsel patients or prescribe, and may assist the registered practitioner.
• Model terms. Gemini API terms say not to submit sensitive or personal information to Unpaid Services, and bar clinical practice and medical advice use. Anthropic's usage policy (effective 12 Nov 2026) treats diagnosis, treatment, medication recommendations and triage priority as high risk, requiring qualified human review and AI disclosure.
• DPDP. Rules were notified on 13 Nov 2025, with core obligations reported for about 13 May 2027, per law-firm summaries. Check the Gazette.

So keep the bot to timings, fees, doctor availability and booking. It should pass symptoms and medical questions to a person, with a human handoff always available.

When an AI chatbot isn't worth it

Don't build one if any of these is true:

• Your answers are few and fixed. The free WhatsApp Business app gives you up to 50 quick replies, greeting and away messages, and labels.
• The problem is slow replies out of hours. A greeting or away message may solve it.
• Most chats need clinical judgement or a staff decision. The bot would only pass them along.
• Nobody will keep timings, fees and schedules current. A confident wrong answer is worse than a late one.
• A fixed menu is enough. Compare a BSP plan before commissioning code.
• Your team relies on the app. Meta supports app-and-API coexistence on one number, with limits such as groups not syncing and broadcast lists being read-only. API messages stay billable.
• The payback is zero or negative.

Here's the payback formula:

Payback months = build quote ÷ (monthly value of staff time saved − monthly running cost − monthly upkeep)

If the bracket is zero or negative, don't build. Only you can supply the hours saved: count and time a week of chats, then multiply by what that staff time costs you. We have no savings figure for your desk.

What Cybiqon would build for this

If the sum works, Cybiqon would write an agent for your front desk's actual questions. It would answer timings, fees and doctor availability from data your staff can edit, take bookings, and hand symptoms and medical questions to a person. It's written for your clinic rather than configured from a template, the founders build it, and you own the code.

Our AI agents start from ₹19,999, scoped on the first call, with a 1–3 week timeline. Running costs like those above are separate, and that starting price doesn't imply a bot of any given scope.

Bring your chat counts and your other quotes to a 30-minute call, or reach us at +91 9250711473 or support@cybiqon.in. If the free app is enough, we'd rather tell you that.

FAQs

How much does an AI chatbot cost per month in India?

As of 1 Nov 2026, direct running costs in our illustrative example are about ₹859–₹1,352 with GST at 300 chats, and ₹3,740–₹8,671 at 3,000 chats. That excludes build, upkeep, staff time, card forex and BSP fees. Sellers' monthly figures run ₹5,000–₹50,000 and may include those.

How much does it cost to build an AI chatbot in India?

Sellers' published estimates, checked 1 Nov 2026, run from ₹25,000 to over ₹1 crore. Their tiers aren't defined alike, so compare what each quote covers.

Are WhatsApp chatbot replies free?

Not all of them. The first 1,000 service messages per number per month are free. After that, each delivered service message costs ₹0.115 before GST, per Meta's rate card effective 1 Oct 2026. Messages patients send you are free.

Can a clinic use an AI chatbot on WhatsApp?

For timings, fees, doctor availability and booking, with a person available, that's the sensible scope. Symptoms and medical questions should go to a person. Whether specific rules apply to your clinic is a question for your lawyer.

Will these chatbot prices hold in 2027?

We don't know, and no 2027 price list exists. The one dated change we can cite is Google's published Gemini 3.8 Flash rate from 1 Jan 2027. Recheck the sources before you budget.

Conclusion

The AI chatbot cost in India is easier to judge once you split it. The build quote is a negotiation about scope. The monthly bill is arithmetic you can do yourself: WhatsApp fees, tokens, a server, and perhaps a platform. As of 1 Nov 2026, those direct lines are small at a few hundred chats, so scope and upkeep deserve most of your questions. Run the payback formula first. If the bracket is positive and you want a second opinion, Cybiqon is happy to go through it with you.