what is blink?

i run multiple ai agents at once and constantly forget what each thread is about.

the scene is usually something like: three claude code windows on different branches, a couple of chatgpt tabs from earlier in the day, a slack thread where i was helping a friend debug, and half a dozen facebook marketplace conversations with people i'm reselling stuff to. by the afternoon i can't tell you what any of them are doing without re-reading the whole history.

every context switch is a tax. re-reading, re-orienting, re-explaining things to the agent i'd already explained an hour ago. across a dozen threads, most of the day goes to reloading state.

blink is a context layer

blink helps people manage multiple agents without getting overwhelmed. it lets you instantly recall and move context across the tools you already use, so the state lives in one place instead of scattered across browser tabs and app windows.

the long-term product is an autocomplete system that predicts what you want to say next based on what you were doing.

shipping the smart clipboard first

the first version is a smart clipboard. you copy something in one app and paste into another, and blink figures out what the right thing to paste actually is, accounting for the source, the destination, and what you were trying to do.

i'm shipping this first because it shares infrastructure with the autocompleter and gets me in front of users faster. the same context-capture and intent-inference layer sits underneath both. one ships in weeks, the other in months.

who it's for

multi-agent power users. engineers, researchers, and operators running parallel ai workflows across tools that don't talk to each other. there's no good answer for them today. that's where blink starts.