First, it's not at all clear that Proto-Indo-European used the augment (temporal or syllabic): it's found in Greek, Indo-Iranian, Armenian, and Phrygian, but not in the other languages, so most likely it was not an obligatory marker of the imperfect and aorist in PIE, if it was used at all.
As for Latin, as far I know, there's no evidence it ever used an augment. (I should maybe say "almost no evidence" because I have a vague memory that there is an archaic inscription in some Italic language which has been argued to contain an augment, but I can't remember the details at the moment, and that interpretation is contested.)
The explanation of ēgī, and other perfects with long ē, is a debated subject; no one to my knowledge has proposed that ēgī contains an augment, partly because this would divorce it from other such verbs where the ē is not initial (e.g. lēgī, sur-rēgī), and it would be preferable to have a unified account of the entire class. (In any case, there is no Attic-Ionic-style change of ā to ē in Latin, so a temporal augment would not yield ēgī.) The theory given in Weiss (2009:413) is that these reflect PIE imperfects of the so-called "Narten class" verbs: these are verbs that show a different ablaut pattern from normal PIE verbs, including a lengthened grade where other verbs have a full grade; in the imperfect, this means such verbs had a long ē.