AI that reads every WhatsApp conversation on 15 criteria
You do not tag anything. Chatinbox scores every WhatsApp conversation on 15 criteria in the background. Customers close to buying, conversations with objections and risky cases surface at once. Personal data is masked without you doing anything.
Chatinbox scores every WhatsApp conversation on 15 criteria in the background. Customers close to buying, conversations with objections and risky cases surface. Personal data such as national ID, IBAN and phone numbers is masked before analysis. You do not tag anything yourself.
The engine runs on Claude Haiku. It reads each conversation on fifteen criteria, including sentiment, intent, buying signal and objection type. It marks for you which conversation is ready to close and which is about to slip away. National ID, IBAN and phone numbers are masked before analysis.
Zeynep Yildiz
15 min · 12 messages
8.2
Satisfaction
91%
Buying intent
LOW
Risk
Numbers flow in real time; a loss shows the moment it happens
Step-by-step conversion tracking and automatic capture of losses. Here is how it works:
15
criteria measured in every conversation
Intent
Buying signals are flagged to help you prioritise
5+
supported languages (TR, EN, AR, DE, FR)
Masking
National ID, IBAN, phone and card details are hidden before analysis
15 analysis criteria
WhatsApp message analysis: every conversation is scored on its own
The Chatinbox Claude Haiku engine quietly runs every incoming conversation through 15 criteria. Your agent adds no tag. Conversations close to a sale, customers at risk of leaving and cases that need escalation rise to the top on their own. You no longer spend hours at a desk reading customer behaviour.
- Sentiment, intent, topic and buying signal come out on their own
- Ready actions for escalation, callback and closing
- Multi-language support (TR, EN, AR, DE, FR)
Personal data is masked without you doing anything
National ID, IBAN, phone, credit card, e-mail... Before analysis starts, the system strips these out. Only the masked version reaches the AI. Personal data does not appear in analysis or reports. When needed, an authorised user views the raw data with one click. Compliance is assessed together with your organisation's own processes and permissions.
- National ID, IBAN, phone, card and e-mail are hidden automatically
- Personal data is masked before analysis; compliance is assessed together with your organisation's processes and permissions
- Raw data viewable by permission
Customer message
Sent to the AI (masked)
Anomaly detection · Sample scenario
1 alert47 refunds in the last 2 hours · Normal: 11/day
Possible cause: quality issue with the campaign product
This week · Ad effect likely
Last 30 days · Score 8.9/10
A trend shifts or something breaks: you hear about it first
The AI keeps watching message volume, topic distribution and shifts in customer attitude. When it spots something unusual, an alert lands at once with a suggested cause. Example: refund requests tripling within two hours.
- Volume, topic and sentiment trends tracked live
- Alerts without delay on sudden spikes
- AI suggestion for the likely cause
Team visibility: who wrote what, and what is still open
AI analysis does not look only at the customer side. It also shows which agent wrote what to which customer, which request stayed open and which one waits for follow-up. A manager finds unanswered requests at the top of the list. Whoever takes over does not have to read the history from the start.
- Which agent wrote what to which customer
- Open and unanswered requests in one list
- Conversations waiting for follow-up are marked
Open requests · Sample scenario
Summary
Two requests are unanswered today; both are stuck on a price question.
Your conversations are data: let AI read them
Have every conversation analysed automatically on 15 AI criteria. Which customer is close to closing, which is about to be lost: base your decision on data.
How to do it
Step-by-step guides from the help center for this feature:
