|
As companies create more applications using AI, corporate officers say demand is rising for a new type of software that stress-tests those apps, simulating how they handle network problems and other disasters that humans may not think to test for—including cyberattacks by rogue AIs.
Luis Chávez, CTO of coding assistant provider Replit, says his company uses an AI-assisted software known as a deterministic simulator, a kind of flight simulator for software. The deterministic part refers to how the simulator lets companies run the same disaster test again and again so they can try out different fixes for bugs that crop up during network failures or hacking attempts.
Traditional stress-testing methods rely on developers guessing every potential problem that may arise. Merely using today’s AI models to evaluate code often produces false alarms or fake warnings, corporate executives say.
Another issue is that AI models behave in a probabilistic manner, so they often won’t act the same way twice and are “very hard” to use to find all the problems in software, Chávez said.
Some companies like Lumen and Pegasystems say they have developed their own software testing tools using AI products from Anthropic and Microsoft. Others are using software developed by startups focusing on the problem. Replit, for instance, uses one from Antithesis, a 200-person, Virginia-based startup. Founded in 2018, Antithesis’ other customers include Palantir and Datadog, according to its website.
Antithesis’ simulator is powered in part by AI it developed in-house, said CEO Will Wilson.
AI made writing software a lot simpler, but “checking to make sure the software actually works…did not get easier,” said Wilson.
Other startups such as Gremlin and Chaos Mesh—not to mention big tech companies Microsoft and Amazon—sell software that tests how companies’ code handles potential problems through controlled experiments on their actual servers rather than in simulated environments. Microsoft is seeing an increase in customers using its testing software, according to a spokesperson.
Wilson thinks Anthropic and other AI coding firms like SpaceX’s Cursor will eventually launch deterministic simulators. (Representatives of those firms didn’t immediately respond to a request for comment.)
Wilson said Antithesis added more revenue in the August quarter than in all of the preceding five periods, though he declined to provide figures. Antithesis raised $105 million in a Series A round at the end of 2025 led by quant fund Jane Street at a $525 million post-money valuation. It’s likely to raise capital again soon, seeking a considerably higher valuation, according to a person familiar with the firm.
|
Gift Subscription