Avatar AI Assistant for small business

The Avatar AI Assistant is responsive and works perfect in small devices

Inspiration

I have been unemployed since January 16, 2026, and I decided it was the right time to get up to date with AI, either to find a new job or to offer AI development services to prospective clients.

A friend of mine owns a small financial services office where they prepare tax returns and sell auto insurance, primarily serving the Latino community in my area. That became my inspiration. I wanted to help him improve customer service by building an AI assistant chatbot. In fact, just last week I visited his office and gave him a link so he could test the chatbot by asking questions related to his business.

When I got home that day, I noticed an email about OpenAI Build Week. It was already Friday, July 17, and I realized I was four days late! I had just discovered the competition, so I spent the weekend rushing to add an interactive avatar to the project.

My friend received the first version without the avatar—it only included the chatbot. The idea of adding an avatar came to me almost by accident after I came across https://face-bot.ai/. I thought, "It would be really cool to combine an avatar with the chatbot!"

Without Codex, I don't think I could have built and integrated the avatar that quickly over a single weekend. It helped me turn an idea into a working prototype in just a couple of days.

What it does

Customers often have questions about taxes, insurance, business services, required documents, and appointments. But finding the right answer can mean searching through several pages or waiting to speak with someone. For OpenAI Build Week, I created the Avatar AI Assistant: a bilingual, customer-facing assistant that turns trusted company information into clear, actionable answers.

How we built it

I built it using ChatGPT to create the prompts and Codex with GPT-5.6 Sol to write the code. I have 25 years of experience in IT, developing software by following the Software Development Life Cycle (SDLC) methodology. I'm not blindly "vibe coding"—I review, understand, and thoroughly test the code throughout the development process to ensure it works as intended.

Challenges we ran into

When I first started learning AI back in February, I was confused about which model and platform to use. I tried Claude (Anthropic) and Gemini (through Antigravity), but I'm glad I ultimately decided to stick with OpenAI and ChatGPT. That decision has given me a consistent development environment and the tools I needed to turn my ideas into working applications.

Accomplishments that we're proud of

I’m incredibly proud of how the Avatar AI Assistant is evolving. It has exceeded my original expectations, and I already have a long list of ideas for new features and improvements. Submitting this project is just the first milestone—I’m excited to continue expanding and refining it in the weeks ahead.

What we learned

This project has significantly expanded my skills as a developer. I learned how to build applications with Next.js and TypeScript, work with Supabase databases, and deploy production-ready applications using Netlify with GitHub integration. It has been an invaluable hands-on learning experience in modern full-stack development.

What's Next for the Avatar AI Assistant for Small Businesses

The current version is just the foundation. Here are the next features I plan to implement:

Secure Client Portal – Create a secure client portal with authentication and usage tracking stored in the database.
Secure Document Uploads – Allow clients to securely upload tax, insurance, and other business documents through the client portal.
Customer Feedback & Reviews – Capture user feedback and store ratings and reviews in dedicated user and review database tables to help improve the AI assistant over time.
Appointment Scheduling – Integrate Calendly so clients can schedule appointments directly from the AI assistant.
Lead Capture – Collect contact information from prospective clients and securely store it for follow-up.
Automated Follow-up Emails – Send personalized follow-up emails after conversations to provide additional information, confirm appointments, or nurture new leads.

Built With

css
html5
netlify
next.js
tailwind
typescript