Auto-translated
I replayed the opening again with Haghesh – he entered the battle on the third day with his first hero (Haghesh’s troops: 85 centaurs, 15 orcs, 25 goblins, 1 druid; after the battle – 70 centaurs, 3 orcs, 1 druid – the core was preserved. During the battle, the AI attacked the orcs and goblins twice with a ring spell, and once with a spell on the centaurs). The second AI hero arrived in my case in the middle of the second week, but he was virtually unkillable. I distracted him with secondary heroes, Haghesh took the gold mines, the crown with +6 knowledge, and quietly waited for the main city to develop (I wasn’t lucky with the combat spells this time – there was no terrifying roar, and it helps a lot). I took the city with the Tear of Ash when the computer moved troops out of it, after which the AI completely ignored this city. As a result, at the end of the second week, the gold income was 13,000 per day. I didn’t continue playing. The main idea is not to let the AI capture the orcish Castle, distracting the initially unkillable AI heroes with expendable heroes running around the map. After building up an army, you can combine it with the elf army from the city with the Tear of Ash, buying artifacts (there’s a shop nearby, and you have plenty of money), and ideally getting the terrifying roar and horde fury spells, and then you can safely enter into battle, but it’s better to do it through the city’s defenses.