The Numbers First
When Arjun Sharma says he makes ₹100,000 a month, the first thing worth clarifying is what that actually means. In USD, it converts to roughly $1,200. In the Indian market, specifically for a solo developer in a mid-tier city, it represents a strong full-time income , well above what many Indian tech company employees earn at junior and mid-levels.
He reaches that figure with 5 to 8 clients. Each pays between ₹12,000 and ₹20,000 per month in maintenance fees, plus a one-time setup payment when the project begins. The math is simple. The path to those clients was not simple, and it took longer than the headline implies.
This is an honest account of what he built, the specific decisions he made, and what he believes other people can realistically replicate. His conclusion: the technical part is the easy part. Everything else takes longer and matters more.
Why WhatsApp, Specifically
Arjun doesn't build general AI chatbots. He builds WhatsApp automation systems for small Indian businesses. The choice of WhatsApp is not a preference , it's an observation about where Indian business actually happens.
WhatsApp is not a messaging app in India the way iMessage is in the US or Slack is for office workers. It is the primary channel for business communication, full stop. Over 400 million Indian users. Every small business , a boutique, a dental clinic, a coaching center, a local delivery service, a CA practice , already receives most of its customer queries via WhatsApp. The owner or someone they pay sits there through the day answering variations of the same questions repeatedly.
That is a solvable problem, and Arjun solves it. The AI agent he builds handles customer queries, qualifies leads, and books appointments through the WhatsApp interface the customer already knows and uses. No new app to download. No portal to register on. The customer messages the number they already have, gets an answer immediately, and often cannot tell whether it's a person or an agent.
The business owner sees a reduction of 60 to 80 percent in time spent on routine customer messages. That ROI is visible within the first week of deployment, which makes client retention straightforward and referrals frequent.
The Technical Stack
The core stack: WhatsApp Business API for the messaging layer, a Python backend running FastAPI for the application logic, and an LLM API as the intelligence layer. He connects the system to the client's existing CRM or booking software, which varies by business but usually involves a straightforward API integration or webhook.
For the LLM, he started with OpenAI's API. He switched to DeepSeek after evaluating the cost difference. That single switch reduced his API costs by roughly 90 percent with no meaningful quality drop for the structured, context-bounded customer service conversations his systems handle. Cost reduction at that scale changes the unit economics of the business substantially , it's part of why the monthly maintenance fee works as a business model rather than just a one-time build.
The infrastructure is deliberately minimal. He is not building a product. He is building a custom solution for each client, keeping the stack simple enough that he can maintain eight clients' systems without those systems consuming all of his time. Complexity in the stack is a liability when you are the only person responsible for keeping everything running.
The entire technical system took him about two months to get comfortable with. WhatsApp Business API documentation is solid. FastAPI is well-documented and approachable. The LLM integration adds a layer of prompt engineering, which is where the work actually lives. Two months to functional first build is a realistic timeline for a developer who is actively learning.
The Non-Technical Work That Actually Matters
Before Arjun writes a single line of code for a client, he spends significant time documenting their conversation patterns. What questions do customers ask most frequently? What answers never change? What situations require immediate human involvement? What tone and formality level does the business want to project? What does a bad customer interaction look like, and how should the system handle it?
That discovery process , mapping the full range of customer conversations, identifying edge cases, designing the escalation logic , takes longer than the actual build. Often significantly longer. And it determines whether the deployed system actually serves the client's customers well or just technically processes messages.
He describes it as requiring empathy as much as technical skill. You are designing a customer experience. The business owner has spent years cultivating relationships with their customers through personal service. Deploying something that handles messages poorly, or that misses the tone the business has established, damages those relationships. Understanding that weight is part of the job.
Small business owners who don't understand technology , and most of Arjun's clients are in that category , need to trust you before they trust your system with their customer relationships. Building that trust requires communication that meets them where they are, not technical demonstrations that impress other developers. That skill, he says, is what most technically capable developers he talks to haven't spent time developing.
The Pricing Model and Client Acquisition
The fee structure has two components. A one-time setup fee between ₹15,000 and ₹30,000, depending on the complexity of the conversation flows, the number of integrations required, and the amount of discovery and documentation work the project involves. Then a monthly maintenance fee between ₹5,000 and ₹8,000 that covers ongoing monitoring, prompt refinements as the business's needs change, and handling edge cases the system hasn't encountered before.
The monthly retainer is the business. Eight clients at ₹8,000 per month is ₹64,000 in recurring income before any new project setup fees. Two new builds in a month adds another ₹30,000 to ₹60,000 on top. He hits his target reliably without needing high-value clients or complex enterprise deals.
He has never run a paid advertisement. His first three clients came from his personal network , people who knew him and trusted him before they understood what he was building. Each of those clients referred one or two other business owners who faced the same problem. The referral chain built his client list to its current size without any deliberate sales activity beyond doing good work for the clients he had.
He currently limits his client list to 15, not because the technical load would overwhelm him, but because beyond 15 the maintenance time starts crowding out the learning time he considers essential. He turns away new clients regularly. The scarcity is intentional.
What Is Actually Replicable Here
The honest answer is: most of it, but not all of it, and not on the timeline the headline implies.
The technical skills are genuinely learnable. WhatsApp Business API documentation is thorough. FastAPI has a large community and good tutorials. The LLM integration layer is well-covered in public resources. Arjun's estimate of two to three months to a functional first build is plausible for someone with basic development experience who is working on it consistently. That part of the path is clear.
What is harder to replicate is the first client. Arjun had existing relationships that gave him warm introductions to his initial clients. Someone starting without that network has to build it first, which is slower and less predictable than learning a technical skill. Cold outreach to small business owners about AI automation has a low conversion rate unless there is a trusted relationship or referral behind it. The path through a personal network is the path , there isn't a reliable shortcut.
The market opportunity is real. The WhatsApp-dependent small business problem exists in every Indian city. The income numbers are achievable with the fee structure Arjun uses. But the path there runs through trust and relationships, and those take the time they take.
The technical skill is the ticket.
The trust is the business.