Catch a frustrated merchant before they leave a 1-star review.
Convot reads every conversation. It escalates the ones turning hostile, grades every resolved thread automatically, and, when a merchant uninstalls, tells you whether support was the cause. AI built to protect MRR, not just answer tickets.
Free for your first app · No credit card · Import from Crisp in one click
Know when a conversation turns hostile, before the merchant churns.
Every incoming message is read in real time. Most signals are caught instantly, and only the ambiguous ones get a deeper AI read, so it stays fast and accurate without crying wolf. When frustration reaches the escalation threshold, managers get an email, a push notification, and a live alert in the dashboard, with the trajectory and the reason, not just a flag.
- ✓Layered detection, tuned to escalate the right conversations, not every blip
- ✓Frustration levels 0–3 (Calm / Mild / Frustrated / Escalated)
- ✓Manager email + push + in-app Live tab with trajectory + reason
- ✓Good-catch / false-alarm feedback loop for accuracy tuning
- ✓15-minute cooldown; auto-clears when resolved; sidebar pill for all roles
Every conversation graded automatically. No spreadsheets.
When a conversation resolves, Convot runs it through Claude, scoring resolution quality, accuracy, tone, and responsiveness on a 0–100 scale. It tags the topic and intent on the same call (no extra cost), quotes the evidence, and flags the ones that need attention. Weekly report emails keep managers informed without digging into the dashboard.
- ✓0–100 quality score with 4 dimensions: resolution / accuracy / tone / responsiveness
- ✓Auto-tagging: topic + intent on the same QA call
- ✓Per-agent score trend + flagged conversation list in the Quality tab
- ✓Daily/weekly grading sweeps; optional Monday report email
- ✓Evidence quote + issue tags on every graded conversation
After every uninstall, AI answers the hardest question.
When a merchant uninstalls, Convot reopens their most recent support conversation and runs AI churn attribution, was support the cause? The verdict comes back clear: support caused it, may have contributed, or wasn't the cause, with a one-paragraph reasoning and a suggested next action. Aggregated across all uninstalls, this becomes the "support-attributable churn %" metric on your Revenue dashboard.
- ✓Per-uninstall AI triage posted as a verdict card in the reopened thread
- ✓A clear verdict: was support the cause, a contributor, or not involved
- ✓A suggested next action whenever support may have contributed
- ✓"Support-attributable churn %" aggregated on the Revenue page
Reply in English. Deliver in Spanish, Japanese, or 26 other languages.
Live Translate detects the visitor's language automatically. Agents read and reply in their own language; the customer sees a translation. Every bubble is cached, so retranslation is instant. Toggle it per-conversation or set outgoing auto-translate to always on. Incoming is never re-translated, your agents always see the original.
- ✓Auto-detect visitor language; one-click translate for agents
- ✓Both directions: incoming (for agents) + outgoing (for visitors)
- ✓29 language pairs, cached per-bubble, near-instant for repeat content
- ✓Manual language override when auto-detect is wrong
- ✓Per-bubble "show original" toggle, never lose the source text
Summaries on every thread, and you see exactly what AI costs you.
When a conversation resolves (≥10 messages), Convot's AI generates a 2–3 sentence summary stored in the contact sidebar timeline, so any agent can get context in seconds without reading 40 messages. Every AI feature (QA, summaries, escalation) is tracked daily by cost and feature, so you're never surprised by an invoice.
- ✓AI TL;DR on resolve for conversations ≥10 messages
- ✓Summary shown inline in the contact sidebar conversation timeline
- ✓Daily AI cost tracking by feature
- ✓Per-org cost view, full visibility, no surprise bills
AI that serves revenue, not just tickets.
Generic AI in support tools does one thing: suggest a reply. Convot's AI is built around a different question, is this merchant frustrated right now? Frustration escalation catches it while the thread is live. Churn attribution tells you whether the last conversation contributed. Auto-QA shows you which agents are producing replies that actually resolve issues. Together, they form a feedback loop that makes your support team measurably better at protecting recurring revenue.
AI works tighter with these pillars.
Revenue intelligence
AI churn attribution feeds directly into your MRR dashboard as the support-attributable churn % metric.
Explore revenue →Shared inbox
Frustration level pill and AI summary visible in every conversation without leaving the inbox.
Explore inbox →Live chat
Live Translate works transparently in the widget, visitors see their language, agents work in theirs.
Explore live chat →Start catching churn signals. Free.
Frustration escalation, auto-QA, and live translation are all active from day one, no configuration, no per-feature pricing.