In this game I failed to see an easy way for my opponent to cut me and escape.
I then resigned since I believed that the game was lost: my groups were very small, I only had territory on the right while white had all corners and a huge territory on bottom left.
After that I discovered that AI estimation was still 60% black. How could it be? What could black do from this position to profit and gain territory?
The bottom left seems to be quiet open. My DDK senses are telling me to reduce it 
@mark5000 started a full AI review.
I agree with flovo. The lower left has a lot of aji. After exchanging f11 for e14, Black has several options for invading or reducing the lower left area.
Well if the ai being approximately pro strength thinks it can still win, that doesn’t mean that a lot of amateur players will be able to win. So I wouldn’t feel too bad 
But is suppose it should be worth trying something in the bottom left just to see how they respond.
I’d say human variations would be handy, given that the ai variations kind of cut off and sometimes there’s just a bunch of exchanges to be used way later on.
I don’t know.
I can’t see a way to live inside that corner and I think that reducing isn’t enough, provided that white has the whole left side + two more corners.
Perhaps I’m just too weak to see something useful for me in that situation.
Suggestions on possible following for black?
I added a variation to the game showing one idea. The position is complicated, and both Black and White have many potential options. That makes it hard to point to any one line. The more important thing was to try.
I posted a couple ideas, assuming we start with the sente exchange F11-E14.
D2 is actually Leela Zero’s recommended move. It is a standard probe, and you can find here some examples of continuations (in a slightly different configuration). Black should be able to either live in the corner or make sabaki on the outside.
N2 is also a key move and black’s privilege. It gives black half a dozen ko threats in the bottom right.