paste it, say it, share it
An email, a voice note, a message to yourself. No new inbox, no new habit to build.
early access, android
SayDo pulls the action and the date out of the message, asks you once, then puts it where it belongs. On your phone. No account. No cloud.
nothing is saved until you say yes.
the actual loop: capture, extract, confirm.
the six-second version
The thought is already in a message. SayDo starts there, instead of asking you to open another app and type it again.
An email, a voice note, a message to yourself. No new inbox, no new habit to build.
An on-device model pulls out the action and the date. Ambiguous bits go to you, not into your calendar.
One confirm screen. The action lands in your list or your calendar. No chat, no feed, no assistant voice.
the honest comparison
SayDo is not a better task manager. It is the step before one. Here is the difference that matters.
stores the sentence.
You still have to reread it, find the action and retype it somewhere useful.
great once the task exists.
But it cannot hear the commitment in a message. You have to do the extraction in your head first.
summarizes in the cloud.
You paste, read, copy and paste again. The result is a conversation, not an action in your day.
turns the message into a decision.
Extracts the action and date, shows its work, and waits for your yes. On-device, with no account.
No cloud round trip. No second inbox. No “assistant” to talk to.
the parts that are not magic
iPhone is next, but not today. The first build is Android.
It runs on your phone, which is the point. It will miss an ambiguous sentence. That is why nothing saves without your confirmation.
You choose what to paste, share or say. SayDo does not get background access to your mail.
No shared workspaces, no cloud sync, no integrations in the first version. It is a personal capture layer.
Your data stays on the device. Backup is a file you control, not a server you hope is safe.
the fair question
You can. It will happily summarise the message. And then it will be trained on that. Then you will copy the summary into a task app, or forget it.
Consumer ChatGPT, default settings. SayDo does not train on your messages.
SayDo is not trying to be a better conversation. It removes the copy and paste between the message and the calendar.
The output is not a paragraph. It is an action with a date, waiting for a yes.
fair questions
Capture and extraction happen on the device. There is no account and no cloud upload in the first build. You can export a backup file you control.
It shows you the action and the date before saving. You can edit, confirm or dismiss it. An uncertain parse is a prompt, not a silent calendar entry.
Yes. The model and the queue are on the phone. Network access is not part of the core capture loop.
No. There is no conversation to maintain. You share a message, approve the result and go back to what you were doing.
After the Android foundation is stable. The waitlist will get the Android build first, and I will say plainly when iOS is ready rather than promise a date now.
why this exists
SayDo started as a small experiment: can a phone pull the commitment out of a message without sending the message to a server? The answer is yes, if the model stays small and the human stays in charge.
built by one person. no growth team, no data business.
the list
The first build goes to this list. I will ask what you want it to catch, because the product should fit the way you already write things down.
join the early access list