Leveraging AI to triage high-volume support channels for County FAs. We combine human-defined logic with AI classification to ensure staff spend less time on administration and more time on the game.
VisualEyes Agent
Deterministic Triage Architecture
Solving the capacity bottleneck in grassroots football governance.
Staff cannot reply late at night. AI doesn't sleep, providing instant answers to routine queries about affiliations or fixtures at 11 PM on a Sunday.
Human advice can vary. Deterministic templates ensure every club receives the exact same, accurate advice regarding FA regulations and sanctions.
By automating 80% of routine "Tier 1" tickets, County FA staff can focus on complex safeguarding cases and strategic development initiatives.
Safe by design. The AI categorizes; the Human dictates the response.
Staff create response templates for specific scenarios (e.g., "Portal Error", "Course Refund").
Inquiry arrives via email or webchat. System strips PII and prepares text for analysis.
LLM analyzes intent. It does not write content. It simply tags:
Intent: Portal_Error
System retrieves the human-written template for "Portal Error" and sends it automatically.
Experience the fully automated flow. No human intervention required for routine queries.
Awaiting inbound message
The demo will start once a new query arrives.
While the action is automated, the oversight is total. Admins have a bird's-eye view of every interaction.
See which templates result in solved tickets and which lead to follow-ups. A/B test language based on real outcomes.
Every AI decision is logged. See exactly why an email was classified as "Refund" and which template version was sent.
Set confidence thresholds. If the AI is only 60% sure, it automatically routes to a human instead of auto-sending.
Total Processed (Today)
1,248
Avg. CSAT
4.8
| Time | Intent | Action |
|---|---|---|
| 10:42 AM | Sanction_Query | Auto-Sent |
| 10:38 AM | Affiliation_Err | Auto-Sent |
| 10:15 AM | Unknown | Flagged |
The system gets smarter, but the human stays in control.
Our system continually refines its accuracy not by "AI self-learning" in a black box, but through explicit human feedback loops and CSAT (Customer Satisfaction) scoring.
Every automated reply asks: "Did this solve your problem?"
If Template #042 drops below 4.0/5 CSAT, it is automatically flagged for review.
Staff update the template logic. The system improves immediately.
"Per regulation 4.2, you are required to..."
"We understand this regulation can be tricky. Here is a simple guide..."
Built for governance, not just speed.
By relying strictly on pre-approved templates, the system cannot invent regulations or give incorrect advice.
Classification models do not train off incoming emails and are fully GDPR compliant, protecting your customers personal data.
Every interaction, classification decision, and template version sent is permanently logged for compliance.
Angry or high-priority emails are detected instantly and bypass automation, routing directly to senior staff.