Build Autonomous Verification Agent

I am looking for an end-to-end autonomous AI agent that can take my manual, multi-location document-verification process and run it hands-free, 24/7. The flow I need reproduced in code is already proven; I simply want it executed without human touch.

Here is how the current workflow operates and what the finished agent must replicate.

1. Extract
My primary GoDaddy mailbox receives new-lead emails around the clock. Each message contains one or more unique IDs plus instructions describing which customer fields are required. The agent has to watch that inbox continuously, detect a new lead the moment it lands, and parse out the IDs and the field list from the body of the email.

2. Scrape
Using those IDs, the agent must authenticate to our secure customer portal (credentials supplied at project start) and collect the exact details requested in the originating email—sometimes just contact information, other times transaction history, account status, or any combination the message specifies. Flexibility here is critical; the data set can change per email.

3. Sync
All scraped data must appear instantly on our internal LMS dashboard so my seven-member team is always looking at one live, authoritative database. I will give you API or database access to the LMS we already run; your job is to push each new record or update without overwriting concurrent edits.

4. Dispatch & Alert
Once the data is in the LMS, the agent should format a concise task brief and send it straight to our pre-approved WhatsApp vendor list. If the email was flagged urgent, I need a real-time alert—push notification, SMS, or another fast channel you propose.

5. Self-Learning
Every time I correct a field mapping or modify an instruction, the agent must remember and apply that preference to future jobs, steadily eliminating repeat errors.

Deliverables
• Production-ready AI agent (Python preferred, but open to other stacks)
• Secure, fault-tolerant email listener for GoDaddy IMAP/SMTP
• Web-scraping module with session handling and captcha bypass if required
• LMS sync middleware with rollback protection
• WhatsApp task dispatcher plus urgent-action alert mechanism
• Simple web or CLI console to view logs, retrain rules, and pause/resume services
• Deployment scripts and handover documentation

Acceptance criteria will be a live demonstration of the four workflow steps completing end-to-end with at least three sample leads, followed by a one-week test where the agent runs unattended and logs zero missed or incorrect tasks.

If you have built similar autonomous or RPA systems and are comfortable integrating with GoDaddy, web portals, and WhatsApp Business API, I’d like to see your plan and timeline.