As of mid-2026, a no-code chatbot subscription typically runs ₹2,000–25,000 a month, while a custom AI chatbot build costs ₹1.5–12 lakh plus ongoing LLM and hosting usage. Channels, integrations, languages, and knowledge-base depth drive the price, and model usage is billed separately on top. These are ranges, not quotes — get your use case scoped before budgeting.
What does an AI chatbot actually cost in India in 2026?
There are two very different price structures, and choosing between them is the real decision. Buying a platform means a monthly subscription that scales with conversations or contacts. Building custom means a one-time development cost plus running costs for model usage, hosting, and maintenance. As of mid-2026, subscriptions commonly land at ₹2,000–25,000 a month and custom builds at ₹1.5–12 lakh.
Most businesses overpay in one of two ways: they buy an enterprise platform for a use case a ₹5,000 tool would solve, or they subscribe forever to something that answers 80% of queries badly when a scoped custom build would have answered 95% well. This guide covers what moves the price, what drives ongoing cost, and how to decide between building and buying. For the strategic view on where bots fit at all, see our guide to AI chatbots across WhatsApp, website and Telegram.
Should you buy a no-code platform or build custom?
Buy when your use case is standard and speed matters; build when the bot must know your data, act inside your systems, or carry your brand voice. Platforms win on time-to-launch and predictable pricing. Custom wins on control, unit economics at volume, and anything that requires real integration.
| Approach | Typical cost (mid-2026) | Time to launch | Best for |
|---|---|---|---|
| No-code platform | ₹2,000–25,000/mo, usage-tiered | Days to 2 weeks | FAQs, lead capture, standard support flows |
| Platform + custom work | ₹75,000–3 lakh setup + subscription | 2–5 weeks | Standard bot with CRM or order-status hooks |
| Custom LLM build | ₹1.5–6 lakh + usage costs | 4–10 weeks | Company-specific knowledge, brand voice, own data |
| Custom agentic system | ₹6–12 lakh+ + usage costs | 3–6 months | Bots that take actions: bookings, orders, refunds, tickets |
Ranges are indicative for Indian development as of mid-2026 and move with scope and seniority.
What actually drives the price of a chatbot project?
Six variables explain almost every quote difference. Understanding them lets you cut cost deliberately instead of just asking for a discount — and lets you spot which vendor has understood your requirement.
- Channels — website widget only is cheapest. WhatsApp, Instagram, Telegram, and in-app each add integration and testing work.
- Conversation volume — drives both subscription tiers and model usage costs.
- Integrations — CRM, order management, ticketing, calendars, ERP. Each is real engineering, and read-write access costs more than read-only.
- Languages — English-only is the baseline; Hindi, Punjabi, or Hinglish support means more testing, more prompt work, and more content.
- Knowledge depth — a 20-question FAQ is trivial; a searchable knowledge base over hundreds of documents is a project.
- Human handoff — routing to a live agent, with context, during business hours, plus the agent console, is often underestimated.
What does the LLM usage itself cost to run?
Model usage is billed per token — roughly, per unit of text in and out — so your running bill scales with conversation volume and how much context each answer carries. For most SME chatbots this is a modest monthly line item; for high-volume consumer support it becomes the dominant running cost and deserves active management.
Three levers control it. First, model choice: smaller, faster models cost a fraction of frontier models and handle routine queries perfectly well, so route by complexity rather than sending everything to the biggest model. Second, context discipline: stuffing entire documents into every prompt is the most common cause of surprise bills — retrieve only the relevant passages. Third, caching and deflection: answer repeat questions from cached responses or static content instead of generating them fresh. Model pricing changes frequently, so verify current rates with your provider and set hard spending alerts on day one.
What are the WhatsApp-specific costs to plan for?
WhatsApp is the highest-value channel in India and the one with an extra cost layer. Beyond your bot, WhatsApp business messaging is billed by Meta through a conversation-based model, typically administered via a Business Solution Provider who adds their own platform fee. Rates differ by conversation category and change periodically — check current pricing with Meta and your provider before modelling.
- Business Solution Provider fee — monthly platform charge, sometimes plus per-message markup.
- Conversation charges — vary by category; user-initiated service conversations are generally cheaper than business-initiated marketing.
- Template approval — business-initiated messages need pre-approved templates; drafting and revising them is real work.
- Business verification — Meta Business verification and display-name approval take time and documentation.
- Opt-in management — you need consent records and a working opt-out, both for policy and for trust.
- Number strategy — a dedicated number, migration planning, and a green-tick application if you want it.
How much does a RAG knowledge base add?
Retrieval-augmented generation — grounding answers in your own documents — typically adds ₹75,000–3 lakh to a build, depending on how messy your content is. RAG is what turns a generic assistant into one that answers accurately about your products, policies, and pricing, and it is usually the difference between a bot people use and one they abandon.
The cost sits mostly in data preparation, not in the retrieval code. Documents must be collected, cleaned, chunked sensibly, and kept current; conflicting or outdated policies must be reconciled before they become confidently wrong answers. Then there is the vector database (modest monthly cost), the embedding generation (a small one-time and incremental cost), and evaluation — a test set of real questions with correct answers, so you can measure accuracy rather than assume it. Budget for a refresh cycle too: a knowledge base nobody updates decays into a liability within months.
What are the ongoing costs after launch?
Plan 15–25% of build cost per year for a custom bot, plus usage. Chatbots are not set-and-forget: your products change, your policies change, and users find phrasings your prompts never anticipated. The bots that keep working are the ones somebody reviews weekly.
- Model and API usage — scales with conversations; monitor and alert.
- Hosting and vector database — modest but continuous.
- Channel fees — WhatsApp conversation charges and BSP platform fees.
- Knowledge-base maintenance — new products, revised policies, seasonal changes.
- Prompt and flow tuning — reviewing failed conversations and fixing them.
- Human agents — someone still handles escalations; the bot reduces that load, it does not delete it.
- Model migrations — providers deprecate versions; plan for periodic re-testing.
| Running cost | No-code platform | Custom build |
|---|---|---|
| Core software | Subscription tier, rises with contacts or conversations | None — you own the code |
| Model usage | Often bundled into the tier, with caps | Billed per token; you control model choice |
| Hosting | Included | Servers plus vector database |
| Upkeep | Content updates in the vendor's console | 15–25% of build cost per year |
How do you calculate chatbot ROI honestly?
The two credible measures are tickets deflected and hours saved; everything else is decoration. Take your monthly conversation volume, multiply by the share the bot resolves without a human, and multiply that by the fully loaded cost of handling one query with a person. Compare against subscription plus usage plus amortised build cost.
Two additions make the case stronger and are frequently ignored. First, revenue capture: enquiries answered at 11pm that would otherwise have gone unanswered until morning — and often to a competitor. Second, response-time effects: faster first replies lift conversion on lead-generation forms and reduce abandonment on support queries. Be equally honest about the debit side: escalations that the bot mishandles cost more than the query would have, so measure containment quality, not just containment rate. Our free ROAS calculator is a quick way to sanity-check the arithmetic.
Build or buy — how should you decide?
Start with the question your bot must answer best; the answer usually decides the approach for you. If it is "what are your hours and prices", buy. If it is "what is the status of my order and can you change the delivery address", build — or at minimum buy a platform and pay for real integration work.
- Buy if your queries are standard FAQs and lead capture.
- Buy if you need to launch in under two weeks or are testing whether a bot helps at all.
- Buy if you have no technical team and no appetite for maintenance.
- Build if answers depend on your own documents, catalogue, or account data.
- Build if the bot must take actions inside your systems, not just talk.
- Build if data residency, privacy, or compliance rules out a shared platform.
- Build if volume is high enough that subscription tiers exceed a custom build within 18–24 months.
- Start bought, move custom is a perfectly respectable path — validate demand first, then own the stack.
How RioCloud scopes chatbot and automation projects
RioCloud Solutions is a Chandigarh agency founded in 2020, delivering AI automation, chatbots, web and cloud work for 100-plus brands across 12 countries. We scope chatbots backwards from the queries you actually receive: pull a month of real conversations, cluster them, and see how many a bot could genuinely resolve. That number, not a feature list, decides whether you should build, buy, or do neither yet.
Often the honest answer is that a workflow automation solves the problem better than a conversational interface — order updates pushed automatically beat a bot that answers "where is my order". You can see the range of that work on our AI automation services page and on our work page, with longer write-ups in our case studies. If you are comparing the automation layer underneath, our n8n vs Zapier vs Make comparison covers the tooling economics.
Frequently asked questions
- How much does an AI chatbot cost in India in 2026?
- As of mid-2026, no-code chatbot platforms typically cost ₹2,000–25,000 per month depending on conversation volume and channels, while custom AI chatbot development runs ₹1.5–6 lakh for a knowledge-based bot and ₹6–12 lakh or more for an agentic system that takes actions. Usage and hosting costs are additional.
- Is it cheaper to build a custom chatbot or subscribe to a platform?
- Platforms are cheaper for the first year and for standard FAQ or lead-capture use cases. Custom builds become cheaper at high conversation volume, and are the only option when the bot must use your own data, match your brand voice, or take actions inside your systems.
- What are the ongoing costs of running an AI chatbot?
- Expect model and API usage billed per token, hosting and vector database costs, WhatsApp conversation charges if you use that channel, knowledge-base updates, and prompt tuning. For custom builds, budget roughly 15–25% of the original development cost per year for maintenance.
- What does a WhatsApp chatbot cost extra?
- WhatsApp adds a Business Solution Provider platform fee plus Meta's conversation-based messaging charges, which vary by conversation category. You also need business verification, pre-approved message templates, and opt-in management. Verify current rates with Meta and your provider before budgeting.
- How do I measure whether a chatbot is worth the money?
- Multiply monthly conversations by the share resolved without a human, then by your fully loaded cost per human-handled query. Compare that against subscription, usage, and amortised build cost. Add after-hours enquiries captured, and subtract the cost of escalations the bot handled badly.
- Do I need a RAG knowledge base for my chatbot?
- You need one if answers depend on your own documents, policies, catalogue, or pricing. It typically adds ₹75,000–3 lakh, mostly in preparing and cleaning content rather than in code. Without it, a general model will answer confidently and sometimes wrongly about your business.
Next steps
Do the cheap diagnostic first: export one month of customer conversations, group them by question type, and count how many a bot could resolve end to end. If that number is small, fix your content or your workflows instead. If it is large, you now have both a business case and a scope.
Want that analysis done properly, with a build-versus-buy recommendation attached? Talk to us — including when the honest answer is that you do not need a chatbot yet.
