I was doing business development in web3, which in practice means you live in Telegram. Deals happen in DMs, and the bigger ones happen in group chats with people from both sides sitting in them. Nobody signs anything in an inbox. There was no system anywhere. Just chats.
The thing I remember is a Monday call. Someone asked what we had in the pipeline, and the honest answer was that we would have to go read our own messages to find out. So we did.
One deal had been sitting untouched for three weeks.
Nobody dropped it
That is the part worth sitting with. Nobody had dropped that deal deliberately. There was no lapse, no missed reminder, no argument about whose turn it was. It wasn’t anywhere to be dropped from.
It lived in one conversation, on one person’s account, in one person’s head. If the person who owned that conversation was busy that week, or on a plane, or eventually gone, the deal went with them. Not because anyone was careless. Because nothing was holding it.
That is a different problem from “we need to be more organised”. You cannot fix it with discipline, and we tried. A team that sells in chat can be completely diligent and still lose deals this way, because the diligence has nowhere to land.
What already existed
Before building anything, I went looking. Three shapes of tool, and each of them solves something real.
Shared inboxes came closest. They are genuinely good at making a pile of messages manageable, and if your problem is volume, they solve it. But an inbox is organised around what arrived recently. It forgets a conversation the moment it goes quiet, and quiet is exactly when a sales deal needs remembering. A thread that has not moved in three weeks should be the loudest thing on your screen. In an inbox it is the thing you have to scroll to find.
Bot-based tools I could not use at all. A bot is a separate account, and it only knows about messages that get sent to it. My existing chats, my groups, years of history: none of it was there. Worse, the arrangement asks your customers to go start a fresh conversation with a robot before anything counts. They will not do that, and it is unreasonable to ask.
Bulk senders were the third shape. Mailshot tools with a CRM label on the box. They send well. They do not remember who replied, what was agreed, or whose deal it is, which is the entire job.
What I actually wanted was boring
Not clever. The chat itself as the record, tied to a contact, sitting on a pipeline with an owner and a next step. History that stays put when people leave. Telegram treated as a real sales channel inside a CRM, rather than a CRM squeezed into the shape of an inbox.
Concretely, four things:
- The conversation is the record. Not a summary someone pasted in, not a forwarded screenshot. The thread itself, attached to the contact and the deal.
- Deals with a value and an owner. So “what is in the pipeline” has a number for an answer, and the answer does not require reading anything.
- Per person tracking inside groups. Telegram deals live in group chats with people from two companies in them. You need to know who said what, not one collapsed row for the group.
- Memory that survives a person. When someone leaves, their deals stay, with the history intact.
None of that is a feature you would put on a billboard. It is just what a sales system is, applied to the place the selling actually happens.
Why nobody had built it
The assumption underneath most of the category is that Telegram is messaging, or worse, that it is support. You can see that assumption in the products: they come out shaped like inboxes, with queues, assignment rules and response timers.
For a support desk that is right. For the teams I was working with it was not even close. In web3, in agencies, in plenty of markets outside the US, Telegram is not one channel alongside email. It is the channel. First contact, the negotiation, the argument about the contract, the complaining afterwards. All of it. And the deals that matter usually sit in a group chat with people from both companies in it.
That is a deal with a value, an owner and a next step. It is not a ticket somebody needs to close by Friday.
The second assumption is more technical, and it is the one that catches teams out when they go shopping. People assume you connect a CRM to Telegram with a bot, because that is how it works nearly everywhere else. It is a reasonable guess and it is wrong here. Connect through the real accounts your team already uses and all of that history just comes over. From the customer’s side, nothing changed at all. We went through that choice in detail in what is a Telegram CRM.
Where this ended up
Cavyro is that system. It mirrors your team’s real Telegram accounts over MTProto, turns chats, groups and channels into contacts, companies and deals, and keeps the history when people move on.
We are early and I would rather say so than quote you a number I do not have. Small internal teams use it day to day. What comes back is requests far more than complaints, which is the good version of early.
If your deals live in chats right now, you already know which Monday call I am talking about. More about why we build this and what we say no to, or try the full plan free for 5 days and point it at your own Telegram.