That's correct. LazSeeker is not exact - it only approximates scaled marks, and then in turn approximates a UAI from that approximation. The aggregate it returns is not the exact scaled aggregate of the student.
It cites aggregates of 468+ as giving a UAI of 100 as a result of adjustments that I made to the conversion equation to allow it to be more accurate for UAIs below 100. I didn't think it was really a problem, considering that, if the calculated UAI is already up in the high 99s, it's impossble to calculate it with a further degree of accuracy.
The UAIs of students at the very top (99+) and towards the lower end of the scale (55-) are more likely to be 'blurred' due to the 'uniqueness' of their position, and the lack of more specific statistics.