AI-Powered Customer Support Website
Budget / Salary$1,500–3,000
TypeFreelance project
LocationRemote
Posted2 hours ago
I need a full-stack website built to deliver an online customer-support service, and at its core will be an AI agent able to handle real interactions. The site itself can be simple in look and feel, but it must be reliable, secure and ready to scale.
Core agent capabilities
• Data analysis on incoming conversations and stored records
• Automatic report generation (daily, weekly, on-demand)
• Real-time answers to customer questions, including FAQ-style queries as well as free-form requests
Technical notes
The stack is flexible—React, Vue or a lightweight alternative on the front end, with Node.js, Python (FastAPI or Django), or another robust back-end framework are all fine as long as speed and maintainability are kept in mind. For the AI layer, you’re welcome to tap into OpenAI, Cohere or a comparable LLM via API, and you may add a vector database such as Pinecone or Milvus if it helps the agent reason over historic tickets. Whatever choices you make, document them clearly in the repo.
Task-based payment model
Work will be broken into milestones (design mock-up, MVP build, AI integration, testing & hand-off). Each completed task is reviewed; once it meets the agreed acceptance criteria, that milestone is released.
Acceptance criteria for each milestone will include
1. Code pushed to a shared repository with clear commit messages
2. A short loom or screen-share video demonstrating the feature in action
3. Basic unit tests where appropriate
4. A concise README update describing how to run or replicate the feature locally
If this flow is clear, let’s discuss timelines and lock in the first milestone.
Core agent capabilities
• Data analysis on incoming conversations and stored records
• Automatic report generation (daily, weekly, on-demand)
• Real-time answers to customer questions, including FAQ-style queries as well as free-form requests
Technical notes
The stack is flexible—React, Vue or a lightweight alternative on the front end, with Node.js, Python (FastAPI or Django), or another robust back-end framework are all fine as long as speed and maintainability are kept in mind. For the AI layer, you’re welcome to tap into OpenAI, Cohere or a comparable LLM via API, and you may add a vector database such as Pinecone or Milvus if it helps the agent reason over historic tickets. Whatever choices you make, document them clearly in the repo.
Task-based payment model
Work will be broken into milestones (design mock-up, MVP build, AI integration, testing & hand-off). Each completed task is reviewed; once it meets the agreed acceptance criteria, that milestone is released.
Acceptance criteria for each milestone will include
1. Code pushed to a shared repository with clear commit messages
2. A short loom or screen-share video demonstrating the feature in action
3. Basic unit tests where appropriate
4. A concise README update describing how to run or replicate the feature locally
If this flow is clear, let’s discuss timelines and lock in the first milestone.
Apply on Freelancer →
Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.