APP IDEA #42 · AI · CHATBOT BUILDER · B2B SAAS

Chatbase — build an AI chatbot for any business in 5 minutes, doing ₹7.27Cr a month

Upload your docs, website, or FAQ — Chatbase trains an AI chatbot on your data in minutes. Embed it on your website, WhatsApp, or Slack. It answers customer questions 24/7 using only your content. ₹7.27Cr/month on TrustMRR. 12,424 active subscriptions. $19.2M all-time revenue. Ranked #7 on TrustMRR. Founded 2023 by Yasser Elsaid.

₹7.27Cr

Last 30 days revenue (Stripe verified on TrustMRR)

12,424

Active paying subscriptions on Stripe

₹19.2M

All-time revenue — $19.2 million total

Ranked #7 on all of TrustMRR by verified revenue

01 / HOW IT WORKS

What the app actually does

Chatbase's insight is that every business has the same problem: customers ask the same 50 questions repeatedly, and either a human answers them one by one (expensive) or the questions go unanswered (lost sales). A chatbot trained on the business's own data — website, docs, FAQ, PDFs — that answers instantly and accurately eliminates both problems. The no-code setup (upload docs, embed widget) made it accessible to non-technical business owners.

Business owner uploads their data sources

Website URL, PDF documents, FAQ pages, Google Docs, Notion pages — anything that contains the information customers ask about. Chatbase crawls and ingests everything automatically.

AI trains a custom chatbot on that data in minutes

The chatbot learns the business's specific information, products, pricing, policies, and FAQs. It only answers from the uploaded data — no hallucination, no made-up answers. If it doesn't know, it says so.

Embed the chatbot anywhere with one line of code

Copy a script tag, paste it into any website. The chatbot appears as a floating widget. Also integrates with WhatsApp, Slack, Zapier, and WordPress. Non-technical owners can set it up without touching code.

Analytics show what customers ask, what converts, and where the bot fails

Dashboard tracks every conversation: most common questions, customer satisfaction scores, conversations that led to a sale, and questions the bot couldn't answer (so you can add that info to the knowledge base).

02 / INDIA POTENTIAL

Does this work as an India play

India has millions of SMBs, D2C brands, edtech platforms, and service businesses that get flooded with repetitive customer questions on WhatsApp, Instagram DMs, and their websites. Most can't afford a customer support team. An AI chatbot trained on their specific business data, with WhatsApp and Hindi support, is something every Indian business owner would pay for — and Chatbase charges in dollars, leaving a massive pricing gap for an India-first alternative.

Indian businesses run on WhatsApp. An AI chatbot that plugs into WhatsApp Business API — trained on the business's products, prices, and policies — and answers customer messages 24/7 in Hindi/English is the single highest-demand feature for Indian SMBs.

Hindi

Chatbase responds in English. An India-first chatbot that natively handles Hindi, Hinglish, and regional languages — and understands Indian product names, pricing formats (₹, lakhs, crores), and cultural context — is an immediate differentiator.

₹1,999/mo

Realistic India-priced entry tier vs Chatbase's $19/month — opens Indian SMBs, coaching businesses, and D2C brands who want AI customer support at a price point that makes sense in India.

D2C

India's D2C brands (skincare, supplements, fashion) get 100-500 WhatsApp messages daily asking about products, shipping, and returns. An AI chatbot handling 80% of these conversations saves them 2-3 support staff — the ROI pitch is immediate and obvious.

03 / THE WEEKEND BUILD

Friday to Sunday, hour by hour

Scope to one thing for v1: upload a website URL or PDF, get a working chatbot widget you can embed. Skip analytics, skip WhatsApp integration, skip multi-source training. One data source in, one chatbot out. That alone is a product.

Friday

Evening · 3 hrs

7–8 PM Set up Next.js + Supabase. Build the data ingestion flow — user pastes a website URL. Use a web scraper (Cheerio or Puppeteer) to crawl the site and extract all text content from every page.

8–9 PM Build the chunking and embedding pipeline — split extracted text into 500-token chunks, embed each chunk using OpenAI Embeddings API, store vectors in Supabase pgvector.

9–10 PM Build the RAG chatbot — on each user question, embed the query, find top-5 matching chunks via cosine similarity, send to Gemini or GPT-4o with a system prompt instructing it to answer ONLY from the provided context.

Saturday

Full day · 7 hrs

Morning Add PDF upload as a second data source — user uploads a PDF, extract text with pdf-parse, chunk and embed alongside the website data. Both sources feed the same knowledge base.

Afternoon Build the embeddable chat widget — a vanilla JS floating button + chat window that loads on any website via a single script tag. The widget calls your API, displays responses in real-time with typing animation.

Evening Add Hindi and Hinglish response support — detect the language of the incoming message and respond in the same language. Test with 10 Hindi questions against an English knowledge base to ensure accurate cross-lingual retrieval.

Sunday

5 hrs

Morning Build the chatbot customisation dashboard — user sets the bot name, welcome message, brand colors, avatar, and suggested starter questions. These render inside the widget.

Afternoon Razorpay integration — Rs1,999/month Starter (1 chatbot, 1,000 messages/month, 1 data source), Rs4,999/month Pro (unlimited chatbots, unlimited messages, multiple data sources, WhatsApp integration).

Evening Test with a real business website — crawl it, build the chatbot, embed on a test page, ask 20 real customer questions, evaluate accuracy. Record the demo.

04 / APP STACK

What you're actually building with

Next.js 14

Frontend + API

Dashboard, widget API, data ingestion, and chatbot response in one framework.

Cheerio or Puppeteer

Web scraping

Crawls the user's website and extracts all text content for the knowledge base.

OpenAI Embeddings API

Vectorisation

Converts text chunks into vectors for semantic search retrieval.

Supabase + pgvector

Vector database

Stores embeddings, user accounts, chatbot configs, and conversation logs.

Gemini or GPT-4o

Response generation

Generates answers from retrieved context — the quality of RAG retrieval determines chatbot quality.

Razorpay

Payments

Rs1,999/month and Rs4,999/month tiers. UPI-first.

05 / WHERE & HOW TO DEPLOY

Going live

Where: Vercel for the app, Supabase for pgvector and database. The widget JS file should be served from a CDN (Vercel Edge or Cloudflare) for fast loading on Indian connections. Main cost is LLM tokens per conversation — model 1,000 messages/month per user at your pricing tier.

Push to GitHub, import into Vercel.

Enable pgvector in Supabase (Database → Extensions → pgvector).

Add env vars: SUPABASE_URL, SUPABASE_KEY, OPENAI_API_KEY, GEMINI_API_KEY, RAZORPAY_KEY.

Test the full pipeline with 3 real business websites before launching.

Serve the widget JS from a CDN for fast load times.

06 / MARKETING & REVENUE

Getting paying users

How to market it

Post the demo reel: "I pasted this restaurant's website URL. 2 minutes later, an AI chatbot that answers every customer question — menu, timings, reservations, pricing — was live on their site. They didn't write a single line of code." Show the widget working live.
Run 3-5 reels/day targeting Indian D2C brands, restaurants, coaching businesses, and clinics — businesses that get 50-500 WhatsApp messages daily with the same questions.
The WhatsApp angle reel: "Your customers message you on WhatsApp asking the same 10 questions daily. This AI reads your entire website and answers them 24/7 — in Hindi." This converts immediately for Indian SMBs.
Cold DM 20 Indian D2C brands on Instagram with a free demo: "I built an AI chatbot trained on your website in 2 minutes — here is what it looks like." Attach a screenshot of their own chatbot. One demo converts.
Partner with Indian Shopify/WooCommerce agencies — they build websites, you add the AI chatbot as an upsell. 30-50% revenue share per client.

Who pays, and why

Indian D2C brands (skincare, supplements, fashion) getting 100-500 WhatsApp messages daily with product questions — an AI chatbot saves them 2-3 support staff.
Indian coaching businesses and edtech platforms where students ask the same curriculum questions repeatedly — the chatbot becomes a 24/7 teaching assistant.
Indian restaurants, clinics, salons, and service businesses that lose bookings because they can't respond to enquiries fast enough — instant AI responses convert more walk-ins.

Scenario

Paying users/mo

Revenue/mo

Slow start

100 × Rs1,999

Rs1,99,900/mo

D2C brand traction + reels

1,000 × Rs1,999

Rs19,99,000/mo

Agency partnerships + WhatsApp tier

5,000 × Rs3,000 avg

Rs1,50,00,000/mo

07 / START BUILDING

Paste this into Claude or GPT

This prompt sets up the full build context so the AI scopes, plans, and starts coding the project with you from message one.

BUILD_PROMPT.txt

o ship a working version in a single weekend. Core product: 1. Business owner pastes their website URL. The system crawls every page and extracts all text content using Cheerio or Puppeteer. 2. Extracted text is chunked into 500-token segments, embedded using OpenAI Embeddings API, and stored in Supabase pgvector. 3. User can also upload PDFs (product catalogs, FAQs, policy documents) as additional data sources. 4. A RAG chatbot is created: on each customer question, embed the query, cosine similarity search for top-5 chunks, generate a response using Gemini or GPT-4o with a system prompt that restricts answers to only the provided context. 5. An embeddable chat widget (vanilla JS, one script tag) is generated — the business owner pastes it into their website and the chatbot appears as a floating button. 6. Hindi and Hinglish support: detect message language, respond in the same language. Cross-lingual retrieval from English knowledge base to Hindi responses. 7. Customisation: bot name, welcome message, brand colors, avatar, suggested starter questions. 8. Conversation analytics: most asked questions, satisfaction scores, unanswered questions. 9. Pricing: Rs1,999/month Starter (1 chatbot, 1,000 messages/month), Rs4,999/month Pro (unlimited chatbots, messages, WhatsApp integration). Stack: Next.js 14, Cheerio or Puppeteer for web scraping, OpenAI Embeddings for vectorisation, Supabase + pgvector for vector storage, Gemini or GPT-4o for response generation, vanilla JS embeddable widget, Razorpay for UPI-first billing. Deploy: Vercel. Help me step by step: 1. Build the website crawler that extracts text from all pages of a given URL. 2. Build the chunking and embedding pipeline with pgvector storage. 3. Build the RAG retrieval and response generation. 4. Add PDF upload and text extraction as a second data source. 5. Build the embeddable vanilla JS chat widget. 6. Add Hindi/Hinglish language detection and response. 7. Build the chatbot customisation dashboard. 8. Build conversation analytics. 9. Wire up Razorpay. India-first: WhatsApp Business API integration is the #1 feature request Indian businesses will have. Hindi support is mandatory. Price at Rs1,999/month to undercut Chatbase ($19/month) while being more accessible to Indian SMBs. If I get stuck, tell me to ask @buildwithkanhaa.

Unlock the full build prompt

Enter your email — takes 5 seconds. Get new guides early.

Please enter a valid email.

I want to build an AI chatbot builder platform for Indian businesses, inspired by Chatbase (chatbase.co — Rs7.27Cr/month on TrustMRR, 12,424 active subscriptions, #7 ranked on TrustMRR, $19.2M all-time revenue), scoped to ship a working version in a single weekend. Core product: 1. Business owner pastes their website URL. The system crawls every page and extracts all text content using Cheerio or Puppeteer. 2. Extracted text is chunked into 500-token segments, embedded using OpenAI Embeddings API, and stored in Supabase pgvector. 3. User can also upload PDFs (product catalogs, FAQs, policy documents) as additional data sources. 4. A RAG chatbot is created: on each customer question, embed the query, cosine similarity search for top-5 chunks, generate a response using Gemini or GPT-4o with a system prompt that restricts answers to only the provided context. 5. An embeddable chat widget (vanilla JS, one script tag) is generated — the business owner pastes it into their website and the chatbot appears as a floating button. 6. Hindi and Hinglish support: detect message language, respond in the same language. Cross-lingual retrieval from English knowledge base to Hindi responses. 7. Customisation: bot name, welcome message, brand colors, avatar, suggested starter questions. 8. Conversation analytics: most asked questions, satisfaction scores, unanswered questions. 9. Pricing: Rs1,999/month Starter (1 chatbot, 1,000 messages/month), Rs4,999/month Pro (unlimited chatbots, messages, WhatsApp integration). Stack: Next.js 14, Cheerio or Puppeteer for web scraping, OpenAI Embeddings for vectorisation, Supabase + pgvector for vector storage, Gemini or GPT-4o for response generation, vanilla JS embeddable widget, Razorpay for UPI-first billing. Deploy: Vercel. Help me step by step: 1. Build the website crawler that extracts text from all pages of a given URL. 2. Build the chunking and embedding pipeline with pgvector storage. 3. Build the RAG retrieval and response generation. 4. Add PDF upload and text extraction as a second data source. 5. Build the embeddable vanilla JS chat widget. 6. Add Hindi/Hinglish language detection and response. 7. Build the chatbot customisation dashboard. 8. Build conversation analytics. 9. Wire up Razorpay. India-first: WhatsApp Business API integration is the #1 feature request Indian businesses will have. Hindi support is mandatory. Price at Rs1,999/month to undercut Chatbase ($19/month) while being more accessible to Indian SMBs. If I get stuck, tell me to ask @buildwithkanhaa.

Start building this weekend

Send me a screenshot of what you ship — it might be the next reel.

DM @buildwithkanhaa →

Want personal help building this?

I personally work with builders on AI, online income & content creation. Run @buildwithkanhaa & @rawyapster.

Work with me →