With the rapid adoption of AI agents in developer workflows, organisations face a critical challenge: there is no standard way to test them. 

These agents interact with users and systems in highly dynamic ways, making it difficult to ensure reliability, accuracy, and consistency. Traditional testing methods fall short when the system under test is inherently unpredictable.

LambdaTest’s Agent-to-Agent Testing introduces a first-of-its-kind solution. The platform employs a suite of specialised AI testing agents to rigorously validate chat and voice-based AI agents across multiple dimensions including conversation flows, intent recognition, tone consistency, reasoning ability, and more.

Teams can upload requirement documents in various formats—text, images, audio, or video—and the system automatically performs multi-modal analysis to generate test scenarios that simulate real-world challenges. 

Each scenario comes with precise validation criteria and expected responses, executed within HyperExecute, LambdaTest’s next-gen test orchestration cloud that delivers up to 70% faster execution compared to traditional automation grids.

The platform also tracks key quality metrics such as bias, completeness, and hallucinations, giving teams deeper insights into the performance of their AI agents. By leveraging a multi-agent approach with multiple large language models (LLMs), LambdaTest ensures broader, more diverse, and more accurate test coverage than single-agent testing systems.

“Every AI agent you deploy is unique, and that’s both its greatest strength and its biggest risk,” said Asad Khan, CEO and co-founder at LambdaTest. “Our Agent-to-Agent Testing platform thinks like a real user, generating context-aware test scenarios that mirror real-world challenges. Each test includes clear validation checkpoints and the responses we’d expect to see.”

Enterprises adopting Agent-to-Agent Testing can expect faster test creation, reduced testing cycles, and up to a 10x increase in test coverage. By automating much of the QA process, organisations can cut costs while improving test depth and reliability.

With 15 purpose-built AI testing agents, ranging from security researchers to compliance validators, LambdaTest’s Agent-to-Agent Testing enables teams to ship AI-powered applications with confidence, ensuring they are robust, safe, and enterprise-ready.

Just six months after its inception, the company is already running pilot programmes in four countries, with paying customers on board.

Co-founders Karim Jouini and Jihed Othmani had already achieved what many African tech founders dream of, selling a startup for a huge sum. Their company, Expensya, was acquired by Swedish procurement software firm Medius in what insiders say was a deal worth over $120 million. 

After that, they swore off launching another company. Jouini even moved into a CTO role at Medius, leading the integration of six firms across three continents. But the itch returned.

“We promised not to do another company because Expensya was too hard,” said Jouini. “But I think it’s like when people have two kids, they forget how hard the first one was. This new venture is less than six months old and already super intense, but we’re fired up. We’re convinced this is unicorn material.”

That fire led to the creation of Thunder Code, a software testing platform powered by generative AI. The product tackles what Jouini describes as the industry’s most annoying bottleneck—slow, manual testing that hinders fast software delivery. 

Their AI agents function like digital QA specialists, identifying bugs, testing interfaces, adapting to changes, and even learning from user feedback.

The idea took root during Jouini’s time at Medius. Watching how development teams struggled with outdated testing tools, he saw a pattern where testing was painful for everyone. That realisation led to Thunder Code’s mission to modernise testing across industries.

The team released its MVP in week six. “Now the product is much more solid six months in than Expensya was in year four,” said Jouini. Speed is a strategy. The startup’s execution playbook is built on lessons from Expensya, don’t overbuild, hire top talent early, and invest in what actually works.

Their approach is straightforward. Write test cases in plain English, let the system convert them to automated tests, use AI personas tailored to different use cases, accessibility, finance, UX, and get actionable feedback immediately. 

That alone has won the attention of QA leads and DevOps teams looking to escape the bloat of legacy testing platforms.

“We’ve been using Thunder Code for our CI/CD pipeline management. The integration with GitHub was surprisingly smooth, and the automated test reporting has made our QA process much more efficient. Saved us countless hours of manual work,” said Mike L., a DevOps Engineer.

Thunder Code’s AI helps with creation and also keeps tests from breaking when apps change. Its auto-healing technology adapts on the fly, reducing test flakiness that usually wastes engineering hours.

The founders are positioning Thunder Code as lean and scalable. Jouini believes AI gives them an edge, not just in product, but in how the company operates. 

“A lot of African entrepreneurs are scared to dilute capital because they want to keep 100%. We believe that if we create a unicorn while diluting ourselves, that’s good value,” he said. He also predicts the company can generate “10 times the value with fewer people” than traditional tech teams.

That lean model has helped them stand out in a market where both legacy players like Tricentis and BrowserStack and new entrants like Nova AI are still defining their next moves. Thunder Code is moving faster and also willing to go broad, with plans to expand beyond web to mobile, desktop, and API testing by late 2025.

Backers from the Expensya days are showing up again. Investors include Silicon Badia, Jaango Capital, Titan Seed Fund, and angels like Roxanne Varza (Station F) and Karim Beguir (Instadeep). 

Even former Expensya employees who profited from the last exit are putting money back in. “Some of our investors are actually Expensya employees and I’m glad it worked out that way,” said Jouini.

Headquartered in Paris with a second office in Tunis, Thunder Code enters a competitive field with real traction, real customers, and a strong sense of purpose. 

For Jouini and Othmani, this is a second attempt at building something far larger than before, faster, and designed for a new phase of software development.