I’m building NewAICodes Chatbots, a platform that helps businesses add an AI assistant to their websites. This post explains the product decisions behind it—especially tenant identity, business knowledge, and simple website integration.
The problem we wanted to solve
A business website may contain useful information, but visitors still have to search several pages, submit a form, or wait for a reply. A website chatbot can shorten that path by answering common questions, explaining services, and guiding visitors toward the next step.
The hard part is not simply placing a chat window on a page. A useful business chatbot needs:
- Knowledge approved by the business
- Instructions and branding for that specific organization
- Persistent configuration and conversation records
- A way to monitor usage and follow up on enquiries
- Straightforward installation on the customer’s website
The basic architecture
At a high level, the workflow is:
- A registered business user creates a chatbot.
- The user adds business knowledge, instructions, appearance, and a welcome message.
- The platform stores that configuration under the user’s account.
- An embed code connects the chatbot to the business website.
- Website visitors interact with the chatbot.
- The registered user manages conversations and usage from the platform.
This keeps the website experience simple while the configuration, knowledge, and management tools remain inside the authenticated application.
Why chatbot creation requires registration
We intentionally do not allow anonymous visitors to create or test a chatbot.
A chatbot is not a disposable text box. It needs an owner because the platform must associate the bot with its configuration, knowledge base, instructions, usage data, and conversation information. Requiring a user identity also makes it possible to keep each customer’s chatbot resources separated and manageable.
The public landing page explains the product, while registered users can access chatbot creation and testing.
For a multi-tenant product, this is an important design principle: every chatbot-related operation should be scoped to the authenticated user and the chatbot they are authorized to manage. That ownership check belongs on the server, not only in the interface.
Business knowledge makes the assistant useful
A general AI model may answer broad questions, but a business chatbot needs company-specific context. The chatbot should be guided by the organization’s own knowledge and operating instructions.
This can help it handle tasks such as:
- Answering frequently asked questions
- Explaining products and services
- Collecting enquiries and potential leads
- Supporting appointment-booking workflows
- Giving consistent, business-approved information
- Preserving conversation history for management and follow-up
The quality of the knowledge supplied by the business remains important. Clear source material and focused instructions usually produce a more useful customer experience than a large collection of unstructured text.
Simple website integration matters
Many small businesses depend on a web designer, freelancer, or agency to maintain their site. Integration therefore needs to fit existing websites without requiring a full rebuild.
NewAICodes provides an embed approach so that a configured chatbot can be added to a website with a small installation step. The business manages the chatbot in the platform, while the website displays it to visitors.
This model can also help web-development agencies offer chatbot setup and ongoing support as an additional service to their existing customers.
Product lessons from the build
A few lessons have stood out during development:
1. Ownership should be designed early
Authentication is not just a login screen. The user, chatbot, knowledge, conversations, and usage records need a clear ownership relationship throughout the data model and API.
2. Configuration should be understandable
Business users should be able to recognize what they are changing: welcome message, knowledge, instructions, appearance, and enabled workflows. Clear controls are more valuable than exposing every possible model parameter.
3. The landing page and application have different jobs
The landing page should explain who the product is for and why it matters. The authenticated application should focus on creating, testing, embedding, and managing the chatbot.
4. Agencies are an important distribution channel
A web agency already understands its customers’ websites and business goals. Giving agencies a simple way to demonstrate and install a chatbot can be more effective than asking every small business owner to become an AI specialist.
Where we are now
NewAICodes Chatbots is available for businesses and web professionals to explore. We are particularly interested in practical feedback from website designers, digital agencies, and small-business teams.
You can learn more here:
If you build websites for clients, I would also be interested to hear which chatbot features make the biggest difference in your projects: support, lead capture, appointment workflows, or something else?
Top comments (0)