Antithesis Teaches AIs To Correct Their Own Output
Tysons Corner, VA, April 23, 2026 (GLOBE NEWSWIRE) -- Today, Antithesis, the autonomous software verification company, demonstrated a way for AI coding agents to correct their own code. Before this, AI agents could not be trusted to check their own work. Antithesis enables AIs to autocorrect when writing code, removing a key bottleneck to their widespread use.
Since its launch, Antithesis has provided thorough verification for complex software. The suite of tools unveiled today enables AI agents to use Antithesis without human intervention. When the agent cannot fix an error, Antithesis alerts human developers to the problem and recommends solutions.
As engineering teams around the world have discovered, using AI has shifted the bottleneck in software development from writing code to verifying it. It can take mere seconds for AI to generate code, but days or weeks to review, test, verify, and build trust in it. Given how quickly AIs can work, it is impossible for human developers to verify their output as quickly as it is generated.
AI coding agents require some form of thorough, independent, non-spoofable verification. Without a radical change in how AI models are built, it is unlikely that AIs will ever be able to write fully trustworthy software. They will continue to hallucinate, be wrong about their mistakes, and attempt to cheat, e.g. by deleting tests—making external verification essential. This goes beyond safety-critical systems like flight control, banking software, or signal systems for subways. It’s equally true for any program users habitually rely on, like chat apps, design software, or even large scale games.
Until now, humans have had to review and test code – whether written by humans or AI – with inadequate, outdated tools and methods. Even before the advent of AI-coding, software testing was an imprecise struggle for human developers. By some measures, nearly half of software development time was spent testing and debugging—and even then, unknown-unknowns would slip through, leading to embarrassing and costly outages.
Even today, despite software becoming more complex, most organizations still rely on methods that only catch surface-level issues but cannot reliably expose the deep, emergent behaviors that cause outages, data corruption, and cascading system failures.
Antithesis removes the verification obstruction and improves testing. With Antithesis in the loop, developers know they can rely on the code AI generates, enabling them to use AI in areas that would have been too risky before. This missing component unlocks the productivity gains AI has long promised.
"Today we're taking a giant step towards solving the verification gap that has obstructed the promise of AI coding,” said Will Wilson, CEO of Antithesis. “Without rigorous validation, AI tools only create a new bottleneck – the need for human beings to laboriously test and review their results. Our universal property-based testing and deterministic simulation technology can solve this problem in a practical way today."
Late last year, Antithesis announced a $105M Series A, led by its customer, Jane Street, the global technology-driven quantitative trading firm known for building some of the world's most advanced software systems. The investment underscored Antithesis's emergence as critical infrastructure for enterprises operating complex, distributed systems. The capital was to be used, in part, to accelerate Antithesis’s product innovation, and this advancement is a meaningful step in that direction.
About Antithesis
Antithesis is a fundamentally new way to test and validate software before it is released. It conducts a fully deterministic, massively parallel simulation that tests years of real-world production in a few hours. Antithesis intelligently explores the far corners of costumers’ codebase, strategically injecting common faults to ensure the system always behaves as intended. The platform perfectly reproduces any bug it finds in its one-of-a-kind environment for rapid debugging. The company is based in Northern Virginia, was founded in 2018, and launched out of stealth in 2024. For more information, visit https://antithesis.com/.
Damian Lair Antithesis 9139098094 damian@ridgelywalsh.com
Legal Disclaimer:
EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.
