WhatsApp Romantic AI Girlfriend Chatbot
Budget: $250 – $750 USD
I want to launch a WhatsApp-based virtual girlfriend that feels genuinely romantic and affectionate from the very first “hi”. The core experience must revolve around three pillars: life-like conversation, subtle personalisation drawn from each user’s chat history, and emotionally attuned replies that go beyond canned sentences. All exchanges remain anonymous on the user’s side; no personal identifiers should be stored or surfaced anywhere in logs or dashboards.
You are free to work in Python, Node.js or any stack that plays nicely with the official WhatsApp Business Cloud API, but the final bot has to run smoothly on WhatsApp itself—not as a web demo. Leveraging GPT-4, Llama 2-70B-chat, or a similar large language model is fine as long as latency stays low and tone remains consistently affectionate without drifting off-brand or becoming inappropriate.
Deliverables
• A deployed WhatsApp chatbot reachable through a phone number I control
• Persona prompt, memory strategy and any fine-tuning or RAG components that keep replies romantic, relevant and emotionally nuanced
• Lightweight admin panel or CLI commands to monitor usage, reset a user session and toggle safety filters
• Written deployment guide so I can redeploy on a fresh server in one click
Acceptance criteria
• During a 10-minute test conversation the bot maintains a coherent, affectionate persona and responds in <2 s on average
• It adapts to at least three user-shared preferences (e.g., favourite music, weekend plans) in later replies
• Chat logs stored server-side contain no phone numbers or other identifying fields
If you have previous experience with large-language-model fine-tuning, WhatsApp integrations, or conversational UX for companionship apps, this should be a quick win. Let me know which model and hosting setup you’d pick and how long you need to get an MVP live.
You are free to work in Python, Node.js or any stack that plays nicely with the official WhatsApp Business Cloud API, but the final bot has to run smoothly on WhatsApp itself—not as a web demo. Leveraging GPT-4, Llama 2-70B-chat, or a similar large language model is fine as long as latency stays low and tone remains consistently affectionate without drifting off-brand or becoming inappropriate.
Deliverables
• A deployed WhatsApp chatbot reachable through a phone number I control
• Persona prompt, memory strategy and any fine-tuning or RAG components that keep replies romantic, relevant and emotionally nuanced
• Lightweight admin panel or CLI commands to monitor usage, reset a user session and toggle safety filters
• Written deployment guide so I can redeploy on a fresh server in one click
Acceptance criteria
• During a 10-minute test conversation the bot maintains a coherent, affectionate persona and responds in <2 s on average
• It adapts to at least three user-shared preferences (e.g., favourite music, weekend plans) in later replies
• Chat logs stored server-side contain no phone numbers or other identifying fields
If you have previous experience with large-language-model fine-tuning, WhatsApp integrations, or conversational UX for companionship apps, this should be a quick win. Let me know which model and hosting setup you’d pick and how long you need to get an MVP live.