sanitized watchlist
Back to proof
Web Agent Monitor
A configurable competitor and watchlist monitor that extracts structured page data, detects meaningful changes, and alerts operators.
Synthetic public view Watchlist change monitor
Config targets
Structured extract
Diff judgment
Operator alert
Fake pagesSQLite historyReviewable alert
LLM-assisted judgment
operator triage
Recruiter answer
What proof does this case study show?
This case study shows AI implementation ability through target configuration, extraction criteria, change-detection judgment, history, alert design, and operator triage.
Role fit
- Customer-facing AI adoption
- SaaS operations workflow monitoring
- LLM-assisted extraction and review
Validation signals
- Separation of fetching, extraction, history, and alert logic
- Human-readable diff reasoning
- Synthetic public targets only
Problem
Competitive and market signals are easy to miss when operators rely on manual checking, ad hoc notes, or noisy alerts.
What I Designed
Specified the monitoring workflow, config model, extraction criteria, diff criteria, storage model, and alerting behavior, then used AI-assisted coding to implement the service.
Validation
The workflow separates page fetching, structured extraction, history, and change judgment so a human can review why an alert fired.
Public Handling
The public version uses fake competitor pages and synthetic alert examples. Real URLs, logs, secrets, and run history stay private.
Synthetic artifacts
What the public proof can show safely.
These artifacts describe the workflow shape without exposing private data, credentials, logs, internal strategy, or real customer/user records.
Configured targets -> structured extraction -> meaningful diff -> operator alert.
Fake competitor change summary with confidence, history, and review action.
Human-readable diff reasoning before a change becomes an operator signal.
No real watched URLs, Slack hooks, logs, databases, or environment files are shown.
Positioning note
This work is framed as AI-assisted implementation. My ownership is problem definition, workflow design, evaluation criteria, validation, rollout, user feedback, and adoption. Coding-heavy pieces were built with AI coding assistance.
Good fit for teams that need AI-assisted monitoring workflows that separate signal from noise and keep operators in control.