Hi there,
My take
Anyone can build an AI that generates words. The hard part, the part that actually matters, is understanding what a person really means. Sarcasm read as sincerity. A furious customer handled like a casual one. A genuine question mistaken for an attack, or an attack waved through as feedback. Get that wrong and the AI doesn’t just miss, it embarrasses you.
That’s the work we’ve been quietly obsessed with. This week Chatti got noticeably better at reading tone: telling sarcasm from sincerity, real opposition from an honest question, and knowing when something is serious enough to hand to a human instead of handling it alone. It’s the same instinct behind ChattiCare removing genuine hate from a public figure’s feed while leaving fair criticism untouched. Judgment, not just words.
And understanding has one more dimension we’re about to crack: language itself. Soon Chatti will read your customer in their language and hand it to you in yours, across more than 90 of them. Because understanding a customer shouldn’t depend on which words, which tone, or which language they happened to use.
Two-minute read. Let’s go.
What’s new this week
- Smarter, safer replies. Chatti’s read on tone just got sharper: it’s better at telling apart sarcasm, genuine questions, and real opposition, instead of taking every message at face value. And high-risk content now automatically escalates to a human rather than being handled on its own. Stronger brand protection exactly where it matters most.
- Greetings that always match your live offer. The first message a visitor sees is now built straight from your agent’s current instructions, so it always reflects what you’re actually running right now, which matters most when you’re promoting something time-sensitive. Greetings are more varied too, so returning visitors don’t get the same opening line twice.
- Every visitor who leaves their details becomes a contact. Chatti now creates a contact even when a visitor leaves their details without starting a conversation first, so someone who reached out is never left unrecorded.
What’s cooking (and it’s big) đź”®
- Automatic translation, in 90+ languages. The big one. Conversations and comments will arrive in your own language, across more than 90 languages including regional variants. Flip on a global preference so everything comes to you translated, or translate any single message on demand, with the original always one click away. It works on every channel, and the language preference belongs to each teammate, since two people on the same team may need different ones. In active development.
- Europe went from intent to an approved plan. Our European expansion is fully scoped and approved: a complete European version of everything we run today, designed so that a European workspace will operate end to end in Europe and no data from it will be stored outside Europe. Built so we can serve European customers with confidence, data residency and GDPR included.
- Chatti Memories moved into testing. Chatti remembering the relevant context about each contact across conversations and channels, so returning customers never have to repeat themselves and your campaigns can draw on that history. Now in QA.
- Tell CommentResponder what to do, in plain English. A new capability in the works: describe, in ordinary language, exactly when Chatti should reply, flag, hide, or delete a comment. No rule builders, no workarounds, just say what you want.
Every one of these came from listening. Hit reply and tell me what would move the needle for your business, and there’s a real chance you’ll see it in a future edition. The more you tell us, the faster the roadmap moves.
See you next Wednesday. ⚡
Jason Tremblay
Co-Founder & CEO, ConversionIQ
Get the next edition in your inbox
lands every week. Product updates, growth, and customer wins, distilled into a five-minute read.