It needs to be universal so it doesn't penalize work. If there's any set limit, it'll create pathologies around it like people not taking up a job because it would lead them to loosing money/time.
Imagine everyone gets half of (Fund Amount - Amount Earned). Let's say the fund amount is 12.5k a year, if you earn 8k you get half of the difference (4.5k/2 or 2.75k). That would put you at 10.75k, closer to the 12.5k amount. If you earned 0, you'd get 12.5k/2 or 6.25k.
This sort of "progressive" guaranteed income would satisfy this incentivization-of-work requirement, while still dishing out a fund to people in need.
It still might not properly incentivize people. If a job is ordinarily worth $16/hr then your effective income for low total hour counts is only $8/hr, which might not be enough to incentivize a person to get the requisite education/training or to put up with unpleasant demands of the job. An employer would have to offer $32/hr to generate $16/hr in value to such an employee.
Wages would not be effected by this implementation, just the amount of government aide disbursed to each person. But you bring up an interesting point that it may not incentivize people appropriately, what might work better?
Any scheme reducing total payout based on other income sources suffers from the same flaw, and it's really just a question of magnitude as a function of income level. I'm not entirely certain that's a bad thing (you'd want a real economist for those kinds of questions), but the impact on incentives is unavoidable.
If you don't reduce total payout based on other income sources, you're describing UBI (or perhaps paying people extra when they earn money, though at first blush that alternative sounds fraud-prone and unstable).
This sort of "progressive" guaranteed income would satisfy this incentivization-of-work requirement, while still dishing out a fund to people in need.