Day 9: Feedback System for AI Suggestions to Improve Accuracy
In Day 8, we built a revision history tracker for AI debugging logs. Today, we’ll add a feedback system so developers can rate AI suggestions as “Helpful” or “Not Helpful”. This enables us to improve the AI prompt quality and monitor AI performance.
Step 1: Add Feedback Columns to Logs Table
Run:
php artisan make:migration add_feedback_to_ai_debug_logs_table
In the generated migration file:
public function up()
{
Schema::table('a_i_debug_logs', function (Blueprint $table) {
$table->boolean('feedback_helpful')->nullable()->after('category');
$table->timestamp('feedback_at')->nullable();
});
}
Run the migration:
php artisan migrate
Step 2: Create Feedback Routes
In routes/web.php:
Route::post('/ai-debugger/{log}/feedback', [AIDebugLogController::class, 'feedback'])->name('ai-debugger.feedback');
Step 3: Add Feedback Form to Dashboard Table
In resources/views/ai-debugger/index.blade.php, add a new column:
<th>Feedback</th>
And inside the row:
<td>
@if ($log->feedback_helpful !== null)
<span class="badge bg-{{ $log->feedback_helpful ? 'success' : 'danger' }}">
{{ $log->feedback_helpful ? 'Helpful' : 'Not Helpful' }}
</span>
@else
<form action="{{ route('ai-debugger.feedback', $log->id) }}" method="POST" class="d-inline">
@csrf
<input type="hidden" name="feedback" value="1">
<button type="submit" class="btn btn-sm btn-outline-success">👍</button>
</form>
<form action="{{ route('ai-debugger.feedback', $log->id) }}" method="POST" class="d-inline">
@csrf
<input type="hidden" name="feedback" value="0">
<button type="submit" class="btn btn-sm btn-outline-danger">👎</button>
</form>
@endif
</td>
Step 4: Add Feedback Logic in Controller
In AIDebugLogController.php:
public function feedback(Request $request, AIDebugLog $log)
{
$request->validate([
'feedback' => 'required|in:0,1'
]);
$log->update([
'feedback_helpful' => $request->input('feedback'),
'feedback_at' => now(),
]);
return redirect()->back()->with('success', 'Feedback recorded.');
}
Step 5: Add Feedback Summary Stats (Optional)
Above the table in the view, you can show:
@php
$total = \App\Models\AIDebugLog::count();
$positive = \App\Models\AIDebugLog::where('feedback_helpful', true)->count();
$negative = \App\Models\AIDebugLog::where('feedback_helpful', false)->count();
@endphp
<div class="mb-3">
<strong>Total Logs:</strong> {{ $total }} |
<strong>Helpful:</strong> {{ $positive }} |
<strong>Not Helpful:</strong> {{ $negative }}
</div>
Step 6: Test Feedback Flow
- Visit
/ai-debugger. - Click 👍 or 👎 on any AI suggestion.
- See feedback saved and badges rendered.
- Try submitting multiple — only one feedback allowed per log.
Next Steps
In Day 10, we’ll wrap up with a production-ready cleanup, add exporting, and outline how to extend this AI debugger into a team-wide error intelligence system.
