Normal Strength Draft classes will still experience randomness of player generation, as well as strong and weak positional groups that year. E.g. A generational QB talent (>90ovr and XF dev) can generate on Normal, but on Very Strong has much greater probability as all QB generation gets a lift (imagine a bell curve distribution of Ovr at a position, and then move the curve to the right).
Weak and Very Weak draft classes act like they're shifting the curve to the left but the draft classes generated will still have some players in the 70-75ovr range at a position but the likelihood of >75 drops significantly.
When using a generator that randomises draft class strength varying per group, the draft class expected range of those groups shifts relative to the strengths or weaknesses. E.g. A very strong OT class will have 10+ Top5, 1st and 1st-2nd LT/RT, a handful of 2-3 and Day3 and the rest UDFA and this will displace the other position groups set to Normal strength (so the LT/RT will occupy more of the 1st round, such as the HB, WR, QB etc will get pushed down rounds and even further displace any group set to Weak or Very Weak). Over time this randomness balances out over multiple seasons thankfully, I have a Very Strong QB class followed by a Weak class using my generator and you can plan your Free Agency and draft/trades accordingly.
You can use non randomised draft class strengths with adjusted XP sliders to offset some of the bias in the draft class generator permanently too. E.g. setting WR to permanently Weak with a 10-20% boost to XP slider means less 70+ WR rookies entering the league and less high picks at those positions but all players develop faster (especially with Dev, mentor or and as a future starter etc). Similarly the QB draft classes can be too strong over time, or the defensive groups too weak compared to NFL trends etc.