Steady Work Behind Chatbots

I spent the first few years of my freelance career trading hours for dollars. It worked, sort of. But I kept noticing a pattern. A client would pay me to build something, I would hand it over, and then the conversation would just stop. No recurring work. No ongoing relationship. Just a clean break and the quiet panic of finding the next project.

The shift came when I stopped thinking about deliverables and started thinking about systems. Not complicated systems. Just things that could run after I walked away.

Chatbots became that thing for me. Not the flashy AI assistants you see in marketing videos. Just simple, well built tools that sit on a website and handle the repetitive questions that business owners answer twenty times a day. What are your hours. Do you offer this service. How do I sign up.

I started offering them two years ago. Now they make up about half of my monthly income without taking half of my time. The setup work is front loaded. Then there is a small recurring fee for maintenance, updates, and the occasional hand holding. It is not passive income, but it is closer to it than anything else I have found.

What follows is how I think about this work now. Not a blueprint exactly. Just what has worked for me.

I tried a few different approaches before landing on something that felt sustainable. The first few chatbots I built were overly complex. I wanted them to do everything. They ended up doing nothing well.

Now I use a combination of tools that let me move fast without cutting corners. The platform itself is usually Voiceflow or Botpress depending on the client's needs. Voiceflow is better for clients who want to see the logic visually. Botpress handles more complex integrations. Both let me self host or use their infrastructure, which matters for the recurring model.

The key was standardizing my approach without making every bot look the same. I built a few templates for common use cases. Customer support triage. Lead qualification. Appointment booking. Each template handles the basic flow, and then I customize the details during setup.

This is where the widgets come in. I started tracking my templates and the time they save. Here is what that looks like.

The templates changed everything. Instead of quoting a wide range of hours, I could give clients a clear number. Setup fee based on the template plus the custom work. Then a monthly retainer for updates, monitoring, and any changes they want down the road.

I landed on this model after undercharging for the first three projects. I was so excited to have someone say yes that I priced the setup too low and ignored the recurring piece entirely.

Now I charge a flat setup fee that covers the template selection, customization, integration, and a thorough testing period. That fee ranges from 1,200 to 3,500 depending on complexity. The maintenance fee is 150 to 400 per month depending on how much oversight the client wants.

The maintenance fee is where the sustainability lives. I monitor the bots once a week, review any failed conversations, make small adjustments to the language model settings, and handle any integration hiccups. Most months it takes two to three hours total. Some months it takes more, but those are usually the months where a client decides to add new features, which I bill separately.

I track the recurring revenue separately from project work. Watching that number grow month over month changed how I think about my business.

I learned that most business owners do not want to think about their chatbot. They just want it to work. That is what the maintenance fee buys. Peace of mind. I handle the small things before they become big things.

I am not a developer by training. I learned enough to be dangerous and then learned more as clients pushed me into new territory.

The core technical stack is simpler than people expect. A platform like Voiceflow or Botpress handles the conversation logic. A vector database stores the knowledge base if the client wants the bot to answer questions about their specific content. Zapier or Make handles the integrations with their CRM, calendar, and email tools. Then a few lines of code to embed the widget on their site.

The AI piece is the language model itself. I use OpenAI or Anthropic depending on the client's privacy requirements. Most small businesses do not need the most advanced model. GPT 3.5 or Claude Haiku is plenty for answering basic questions and qualifying leads. The cost is negligible. A busy chatbot might cost 20 to 50 per month in API calls, which I either pass through or build into the maintenance fee.

What took me the longest to figure out was the knowledge base piece. If a client has a website full of content, you cannot just feed it all to the bot and expect it to answer correctly. The bot will hallucinate. It will invent hours and services that do not exist.

The solution was building a clean knowledge base for each client. I extract their FAQs, their policies, their service descriptions, and structure them in a way the bot can actually use. That is usually the most time consuming part of the setup, but it is also the most important.

Here is the workflow I use now. It took a few failed projects to get here.

The testing phase is non negotiable. I run dozens of test conversations, including the strange ones. What if someone types a curse word. What if they ask about a service the client does not offer. What if they just write the letter A. The bot needs to handle all of it gracefully or the client gets embarrassed.

The projects that went well had one thing in common. The client had clear expectations about what the bot would and would not do.

The projects that went poorly had the opposite. The client wanted the bot to replace a human entirely. They wanted it to handle complex negotiations or make judgment calls that only a person should make. I learned to say no to those projects early.

The best clients are the ones who see the bot as a tool for their team, not a replacement. They want to handle the repetitive questions so their staff can focus on the complex ones. They want to capture leads after hours when no one is in the office. They want to give customers an immediate answer instead of a 24 hour email delay.

I started asking better questions during discovery calls. What is the one question you answer most often. What takes up the most time for your support team. What happens when you miss a lead because you were out of the office. Those questions led to better bots and happier clients.

Here is a quick comparison of how different businesses approach this. I found it helpful to see where the opportunities actually are.

Each type of business requires a slightly different approach. The local retailer needs simple hour and availability answers. The SaaS company needs help desk integration. I keep notes on what works for each category so I am not starting from scratch every time.

The thing I did not expect was how much I enjoy this work. I like building something that quietly solves a problem. I like that the client pays me once to build it and then pays me a little each month to keep it running. It feels like a fair trade.

There is also something satisfying about the conversations themselves. Watching a bot handle a frustrated customer at 2 AM with patience and accuracy. Seeing a lead captured that would have otherwise bounced. Knowing that someone got the answer they needed without waiting on an email.

This model works because it solves a real problem. Business owners are tired of answering the same questions. Customers are tired of waiting for answers. The technology is good enough now to handle the middle ground. Not perfect, but reliable.

I started with one client two years ago. Now I have twelve. The work is steady. The income is predictable. And I spend more time on the creative parts of my business than I used to.

If you are curious about this, I would say start with one client. Someone you already know. Build them something simple. Charge a fair setup fee and a small monthly retainer. See how it feels. See if the maintenance work fits your rhythm.

It might not be for everyone. But for me, it has been the quiet shift that changed everything.

Out Of Home advertising has long been effective but hard to scale—until now. AdQuick makes it simple to plan, deploy, and measure campaigns with the same efficiency and insight you expect from online marketing tools.

Marketers agree: OOH is powerful for brand growth, driving new customers, and reinforcing messaging. AdQuick makes it easy, intuitive, and data-driven—so you can treat real-world campaigns like any other digital channel.

Learn more, visit AdQuick.com