modl.ai runs agents that play a game build the way testers do, hunting for bugs, glitches and performance problems, and it does so without requiring an SDK to be integrated first.
Game QA is unusually well suited to this. The work is enormous in surface area, repetitive in method, and impossible to cover exhaustively by hand: a level has more traversable states than a human team can visit before ship date, and the bugs players find are usually in the states nobody visited.
Pricing is not published and the pricing page 404s. Agents also find a different class of problem than human testers, catching traversal and performance faults while missing whether something is actually fun, which no automation addresses.
