The total budget is roughly 10% of what the DAO made this year. I would argue to have an annual review which would adjust this total budget and drop projects that fall off the total diminished (or increased) budget.
Also while it’s up to the governance to call the stream, I’d implement it differently than what Labs do it, but rather more similar to how sablier and super fluid does: there’s a capped maximum budget allocated for the stream where we deposit a given amount. If the DAO does nothing then eventually that amount will drain out and the streams stop automatically.
The difference between just doing an annual grant giving is that if the new process takes a few extra months then the current service streams are uninterrupted while we vote and the new ones are adjusted seamlessly without skipping a single block.