I feel like |9.f10 was the game-winning move. A more ordinary response like |9.i9 would have let 10.i16 work. That looks like a winning position for Black.
Letting the k12 threat linger, playing |8.i16 first, would have probably been the better move, and still looks like a winning position for Black, though it may still be a bit early to be sure.
The bot suggests |10.h9 threatening a pincer attack for black on E12 or F13, so instead white plays 11.q1112.n1813.d914.f1315.g1216.i16 and black seems to have an advantage.
Letting the k12 threat linger, playing |8.i16 first, would have probably been the better move, and still looks like a winning position for Black, though it may still be a bit early to be sure.
After |9.f10, 10.i16 is broken by 11.g16: 12.g15 13.f14 14.g20 15.i17 16.j19 17.j20 or 14.h20 15.d19. So Black needed to find something else in the lower left corner for move 10, and apparently |10.h18 isn't it. Is there a winning move 10?
H18 was probably black's losing move.