Prediction
Ve reads the flow of your work — messages, meetings, commits — and stages the next tab on the left side of your browser, prompt ready. You just say yes.
Join the waitlistVE reads the thread on your screen and drops a reply that matches it. Click inside the reply field or tap the key to watch it happen.
Press and hold the key, ramble the way you'd say it out loud, then let go. VE pastes a structured prompt straight into the composer.
Hold Fn and ask for what you need. VE finds the file across your connected tools and puts it in the composer with the message.
Work arrives
A teammate asks for the rollout doc in Slack. A meeting hands you research. A PR lands with issues. Ve sees it in the flow.
The tab is waiting
By the time you look, it’s staged on the left rail — context gathered, draft written, the prompt already composed.
You say yes
Approve, redirect or dismiss. Say yes twice and Ve offers to run it as a Routine — it stops waiting to be asked.
No frameworks to memorise. No "act as…" templates to fill in. Just hold a key and think out loud.
Where does the prediction come from?
Does it act without asking me?
What if the prediction is wrong?
Can it predict from meetings I didn't record?
How is this different from an agent I have to prompt?