The promises you made in WhatsApp and nobody wrote down
Your business runs on sentences like “I’ll send the quote tomorrow”. They are not in a CRM or on a list — they are in a thread, and they are forgotten at a rate nobody measures because nobody can see it.
“I’ll send the quote tomorrow.” “Let me check and come back to you this evening.” “We can deliver twenty by Friday.”
Every one of those is a commitment. None of them is in a CRM, on a task list, or anywhere a person will look again. They are in a thread, underneath forty newer messages, and the only thing standing between the promise and the customer is somebody remembering.
Why this is invisible
Businesses measure what they can see. Unanswered messages are visible. Response times are visible. A promise made on Tuesday and quietly not kept on Friday leaves no trace at all — the customer does not usually complain, they just stop replying, and it gets recorded as a lead that went cold.
The damage is not the forgotten quote. It is that the customer learns something about you, and what they learn is that your word on WhatsApp does not mean very much. That is an expensive lesson to teach somebody who was about to buy.
Why the usual fixes do not stick
| Approach | Where it breaks |
|---|---|
| Write it in a CRM | Requires switching apps mid-conversation. Works for two days |
| Star the message | A starred list of 200 messages is not a to-do list |
| A task app | Same context switch, plus retyping what you already wrote |
| Remember | Works until you have more than one conversation |
| A colleague reminds you | Now two people have to remember |
Every one of these asks the person who is mid-conversation to stop and do clerical work. That is a reasonable thing to ask once. It is not a reasonable thing to ask forty times a day, which is why none of them survive contact with a busy week.
The commitment is already written down
Here is the thing that makes this solvable: you already typed it. The promise exists as text, in a thread, with a timestamp and a person attached. Nobody needs to write it down again. It needs to be noticed.
Language models are good at exactly this narrow job. Not at running your business — at reading a sentence and deciding whether it contains a commitment by you, to somebody, with a time attached. “I’ll send it tomorrow” yes. “Tomorrow is Thursday” no. “Can you send it tomorrow?” no — that is a promise being requested, not made.
What a useful version has to get right
- Only your promises. A customer saying “I’ll pay Friday” is worth tracking too, but it is a different list and conflating them makes both useless.
- A real deadline. “Soon” is not actionable. “By Friday”, “this evening”, “before the weekend” resolve to a date, and that date is what makes it appear at the right moment.
- Quiet until it matters. A list that notifies on capture is another notification stream. The only moment that matters is shortly before the deadline.
- Closed by doing it. If sending the quote does not clear the item, the list rots within a week and everybody stops trusting it.
- Wrong sometimes, and cheap to dismiss. It will occasionally catch a turn of phrase. One tap to clear is fine; a workflow to clear is not.
What changes when you can see them
The first week is uncomfortable. Most businesses that turn this on discover a number of outstanding promises they would have guessed was near zero, and it is not near zero.
After that, three things shift. People become more careful about what they promise, because it is now written down where somebody will see it. Handovers get easier — a colleague picking up a conversation can see what was already committed to. And the most common outcome is the least dramatic: a quote goes out on the day it was promised, and the customer notices, and says nothing, and buys.
How WaveSenda does it
Messages you send are scanned for commitments. Each one becomes an item with the promise, who it was to, and when you said — visible in the console, and surfaced before it is due rather than after. Marking it done closes it, and so does sending the thing.
It works alongside the shared inbox, which is where it earns most of its keep: a promise made by whoever was on duty on Tuesday is not lost because somebody else is on duty on Friday.
Common questions
Does this read all my WhatsApp messages?
It reads the messages sent from the numbers you have connected, looking for commitments in what you send. It is the same access the rest of the product already has in order to show you your inbox — nothing new is exposed, and the feature can be switched off entirely.
What if it catches something that is not a promise?
Dismiss it, in one tap. A detector tuned to never produce a false positive would miss most real commitments, so the right trade is to catch generously and make clearing trivial. If dismissing is any harder than that, the list stops being used.
Does it track promises customers make to me?
Those are worth knowing but belong on a separate list — "they said they would pay Friday" and "I said I would deliver Friday" need different follow-ups, and mixing them produces a list that is not really about anything.
Will it work if we speak more than one language?
Commitment detection is a language model reading a sentence, so it is not tied to English, though quality varies by language. Worth testing on a week of your own real messages rather than trusting a claim in an article.
What happens to a promise when the conversation is handed to somebody else?
It stays attached to the conversation rather than to the person who made it, which is the point. The colleague picking it up can see what was already committed to and by whom, instead of finding out when the customer asks.