Quality Overview
DEMOLive view of the bot's daily verdicts
/
Daily QC
—
Volume snapshot
Auto 8:00Verdict mix
TodayLost sales
0Weekly QA digest
Agent review queue
Chat verdicts — approve the bot's work
Chat / codePhoneAgentMessagesLast activityStatusVerdictYour review
Drop the updated workbook here
Rows from a workbook are replaced; anything typed into the dashboard is kept.
Destinations
0 —Bookings by month
—Entries
0DateTypeDestinationTravel datesBooking refPhone
Call QA · KPI scoring
Phone Reviews — calls scored on the official rubric
Scoring off — enable in ScheduleScore specific calls now
Add one call at a time — number, date and agent — then score the whole list at once.
AgentPhoneCall dateLengthScoreStatus
Per-agent quality
Agent scorecards
| Agent | Scored chats | Verdict mix | Mistakes | Quality score | Approved |
|---|
Average phone score per agent, from every scored call — corrected scores included.
| Agent | Scored calls | Avg score | Needs review | Random sample |
|---|---|---|---|---|
| — | ||||
Response times
Minutes the customer waited for a HUMAN agent (the AI assistant replies instantly and is excluded).
| Agent | Chats | Avg first reply | Avg wait | Worst wait |
|---|---|---|---|---|
| — | ||||
Management · monthly KPI
Agent KPI
Score each agent
10 + 30 = 40Calls are worth 10 points and chats 30 — the total is worked out for you. Leave a box empty for an agent you have not scored yet; a month is never closed, so it can be corrected any time.
Management · what the bot was taught
What the bot learned
—Rules the bot learned
0Every rule here is read by both judges on every scoring run, for as long as it stays in this list. Read them and delete any you disagree with — a wrong rule quietly bends every score after it.
| From | Reviewer wrote | Rule the bot learned | Action |
|---|---|---|---|
| — | |||
Comments the bot refused to learn from
0These wordings were refused — none of them is in the bot’s head.
| From | Reviewer wrote | What the bot tried to learn | Why it was refused | Action |
|---|---|---|---|---|
| — | ||||
Ask the conversations
Ask
Conversations
—
Ask about the conversations
The counts come from the scored reports — destinations, services, mistakes, agents. Ask in Arabic or English.
Enter to ask · the numbers below every answer are the ones it used
Learn from corrections
Calibration
Recorded corrections retrain the scorer liveUpload corrected reports
Edit عدد الأخطاء / يحتاج مراجعة in the daily Excel, then drop it here. The bot diffs it against its own verdicts, records every human decision to the ledger (so the next report self-corrects), and flags anything needing a code fix.
Drag & drop the corrected Excel files here
What the bot is learning
Auto—
Recent corrections
| Chat | Bot said | Human | Type | Source / fix | By |
|---|---|---|---|---|---|
| — | |||||
Reviewer comments
Fed to the botEvery comment the agents write is fed to the judge on the next scoring run. Substantive ones get turned into code fixes.
| Chat | Decision | Comment | By | Time |
|---|---|---|---|---|
| — | ||||
Worth a re-check
0Counted mistakes with a specific reason to doubt them. This is an automated second opinion that cannot read quote images — it only asks for a human ruling, it never removes anything. Open a chat and press green to keep the mistake or red to remove it.
| Chat | Agent | Why it is listed | Day |
|---|
Admin · audit log
Activity
Recent activity
| User | Time | Action | Target |
|---|---|---|---|
| — | |||
Everyone on the dashboard
Daily volume · Calls & chats
Volume report
Manual daily metrics · Sales & booking channels
Sales
Enter / edit a day
هوتيل % —Pick a date — existing numbers load automatically for editing. hotels % is computed for you: (Hotels + APP hotels) ÷ TD.
Daily trend
Month vs month
Monthly totals
| Month | Airline | non TD | TD | Farekey | APP | Hotels | هوتيل % | B2B | Wego |
|---|
Days in range
| Date | Airline | non TD | TD | Farekey | APP | Hotels | هوتيل % | B2B | Wego |
|---|
Automations
Scheduled reports
Preferences
Settings
Account—
AppearanceLight or dark cockpit
LanguageInterface language & direction
Data source—
—Report recipientsKhaled · server@tathkarah.com
ActiveBitrix workspacetathkarah.bitrix24.com
LinkedScoring engineTwo-brain judge · Layer-2 verify · deterministic detectors
LiveTeam logins
Leave the password empty to invite — the user sets their own on first login.