XoreCare · consumer complaints & quality intelligence
Three complaints. One incident.
XoreCare brings every consumer complaint — your toll-free line, WhatsApp, Messenger, Instagram, the comments under your Facebook posts, email, web chat and SMS — into one inbox. Then it reads the stream for the quality incident forming inside it, while it is still an incident and not a recall.
Product complaints · last 90 days
Read together · same production run
The problem
Every complaint is handled. The pattern is not.
Consumer complaints arrive in more places every year, and each one gets closed on its own. The signal that matters lives in what they have in common — which is exactly what nothing is looking at.
Eight inboxes, no single view
The toll-free line is one system, WhatsApp another, Facebook comments a third — and nobody is reading them beside each other. A consumer who tries two channels becomes two tickets.
A batch number nothing groups by
General helpdesk software has no concept of a production run, a plant or an SKU. It cannot cluster complaints by batch because there is nothing in its model to cluster on.
The reporting clock already started
For a serious concern the window to notify a regulator runs from when the complaint arrived, not from when somebody noticed the pattern. Finding it in a quarterly review is finding it late.
What makes it different
Signal Radar reads the stream, not the ticket
Three consumers report a leaking pack. Pune, Nashik, Aurangabad — three cities, three agents, three different days. Each complaint is handled properly and each is closed with a replacement. Three tickets, case closed three times over.
Except all three packs came off the same production run. Nobody saw it, because nobody was looking at the three of them together.
XoreCare groups product complaints by batch and by plant-and-defect, scores each cluster for risk, and says what to do about it in a sentence a quality manager can act on or overrule. Two complaints on one batch is a coincidence and is deliberately ignored — a radar that reports every pair is a radar people switch off.
- Clusters by production run, and by a defect recurring at one plant across runs
- Ranks a reportable cluster of three above an ordinary cluster of eight
- Carries the window — three complaints in six hours is not three in six weeks
- Scoped to territory, so a cluster never crosses a boundary a complaint could not
- Recommends an action, not a number: hold the batch and pull retained samples
Channels
Your numbers, your Page, your data
You connect your own WhatsApp Business Account, Facebook Page and Instagram account. XoreCare receives messages and replies on your behalf, under your brand, into a database that belongs to you alone. Nothing is shared between clients and nothing is resold.
Two-way on the WhatsApp Business Platform. Text, images, documents and voice notes — transcribed on arrival, and answered as a voice note if that is easier for the consumer.
Messenger & Instagram
Direct messages to your Facebook Page and Instagram account arrive as conversations, with the sender recognised across channels rather than treated as a new person each time.
Facebook Page comments
A public complaint under one of your posts is still a complaint. It opens a conversation like any other, and the reply goes back to the comment rather than into a void.
Voice
The toll-free line and any direct-dial number. The call is captured with the complaint, and the reference is allocated on the call rather than promised afterwards.
Your published complaints address, threaded properly so a consumer's third email joins the first two instead of starting a third case.
Web chat
A widget on your own site. No third party to approve it, no external script deciding what your visitors see.
SMS
For acknowledgements and status updates where a consumer has no smartphone — still the most reliable way to reach much of India.
Portal & mobile app
A consumer-facing page where somebody can raise a complaint and check its status without calling anybody to ask.
How it works
From a message to a decision
Four steps, and the fourth is the one no general helpdesk reaches.
Arrives
On whichever channel the consumer chose. A reference is allocated inside the same transaction as the record, so a failed save never burns a number and leaves a gap you will one day audit.
Understood
Language detected, sentiment and severity read, personal data redacted before anything reaches a model. Your own knowledge base grounds the suggested reply, so it answers from your policy rather than from a guess.
Routed
Categorised against your master data — brand, SKU, plant, batch, geography — and sent to the right team with an SLA clock that started when the consumer wrote, not when somebody opened it.
Connected
Signal Radar reads the whole stream and surfaces the cluster: same batch, same plant, same run. This is the step that turns a complaint desk into an early-warning system.
Capabilities
Everything a complaint desk actually needs
Grounded AI replies
Suggested replies are drawn from your own knowledge base and cite the article they came from. An agent approves or edits — nothing is sent without a person.
22 Indic languages
Read a complaint in Marathi, reply in Marathi. Translation and speech run on the same self-hosted stack as XoreVoice, so nothing is sent to a third-party model to be understood.
Voice notes, both directions
An inbound WhatsApp voice note is transcribed on arrival. The reply can go back as a voice note in the consumer's own language, which for many is the only way they will read it.
SLA and escalation
Your own resolution commitments per category, on a business calendar. Breaches escalate on a schedule you set, and every step is on the record.
Investigation and CAPA
Sample collection, technician dispatch, root cause and corrective action — attached to the complaint rather than kept in a spreadsheet beside it.
Territory scoping
A salesperson sees their own patch and nothing else, enforced on the server rather than hidden in the interface. Reports, lists and clusters all obey the same boundary.
Privacy & audit
Built for the day somebody asks to be forgotten
India's data protection regime gives consumers rights over the personal data you hold. XoreCare was designed around those rights rather than retrofitted to them.
A database per client
Not a shared table with a tenant column. Your complaints live in their own PostgreSQL database, which is also what makes an export or a deletion a clean operation rather than a careful one.
Crypto-shredding
Every consumer's identifying data is encrypted under a key belonging to that consumer alone. Erasure destroys the key, and the record survives with the person removed from it.
Redaction before inference
Names, numbers and addresses are removed before any text reaches a model, and put back afterwards. The model never sees who complained.
An immutable trail
Who saw what, who changed what, and when — sequenced so the order cannot be argued with. Quality records must not change shape retrospectively.
Architecture
Self-hosted, and honestly so
A modular .NET monolith over PostgreSQL, with a database per client and inference on your own hardware. It runs in your data centre, in your cloud account, or in ours — the choice is a deployment decision, not a licensing one.
See it on your own complaints
XoreCare is going into early access with a small number of consumer brands. The quickest way to know whether it fits is to point it at a month of your existing complaints and see what it finds.