Multi-Channel AI Support Automation

via Freelancer ·

Budget / Salary€250–750
TypeFreelance project
LocationRemote
Posted1 hour ago
I need an end-to-end automation that reads incoming customer messages from email, my website chat widget, and our main social media inboxes, feeds the text through OpenAI (GPT-4 preferred), then sends an appropriate, well-formatted reply back through the same channel. Whenever the model’s confidence drops below an agreed threshold, the entire conversation—including context, metadata, and a short AI-generated summary—must be routed instantly to a human team member so we can take over without losing continuity.

I already run other workflows in both n8n and Make, so the new logic has to live in those platforms, making full use of their webhooks, queues, and error handling. Clean, modular scenarios are important because I will be expanding languages and channels later.

Deliverables
• n8n or Make scenarios (clearly commented) that:
– pull messages from Gmail/IMAP, the chat widget API, and Facebook/Instagram direct messages;
– call the OpenAI endpoint with temperature, system prompts, and token limits that we can tweak;
– push replies back to the correct thread, preserving formatting;
– trigger a human-handoff path when confidence is low, including Slack/Teams alert with conversation link.
• A short README or Loom video showing how to adjust API keys, thresholds, and prompts.

Acceptance criteria
1. A test message on each channel receives an AI reply in less than 30 seconds.
2. For forced low-confidence tests, the handoff alert reaches our team channel every time.
3. All scenarios run error-free for 24 hours in production mode.

As soon as the flows are stable I’ll provide real credentials so we can switch from sandbox to live.
php c programming c# programming software architecture automation api integration openai ai chatbot development
Apply on Freelancer →

Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.