Karim Jouini and Jihed Othmani are not first-time founders learning in public. They have already built and sold a large enterprise software company. Thunder Code is what they chose to build next.
Both founders previously co-founded Expensya, an expense management SaaS company launched in 2014 that scaled across Europe, Africa, and the Middle East. In 2023, Expensya was acquired by Medius, a Swedish spend management company. While the financial terms of the deal were not officially disclosed, it was widely reported as a nine-figure transaction and is consistently cited as one of the largest enterprise software exits involving a company founded in Africa.
After the acquisition, Jouini and Othmani stepped away from operating roles. In interviews, they have described taking time to reflect on what had broken most often while scaling Expensya. One issue kept resurfacing. Software testing was slow, manual, and expensive, even inside well-run engineering teams. Release velocity depended not on how fast developers wrote code, but on how quickly teams could verify that it worked.
Thunder Code was founded in 2024 to address that bottleneck directly.
The company is building an AI-based quality assurance platform designed to automate software testing. Instead of relying on scripted tests or large manual QA teams, Thunder Code uses generative AI to simulate human testers. The system deploys autonomous test agents, each assigned a role such as user interface testing, accessibility checks, or performance validation. These agents interact with applications through their interfaces and APIs, identify failures, and report defects without requiring test scripts or code from users.
The first prototype was built in approximately six weeks and launched as a cloud-based service focused initially on web applications. Users describe test scenarios in plain language. The platform translates those instructions into continuous testing workflows. Expansion into mobile, desktop, and API testing is planned as the product matures.
The problem Thunder Code is addressing is well known inside engineering organizations. Quality assurance remains one of the least automated parts of the software lifecycle. Despite modern DevOps practices, a significant share of testing is still performed manually. This creates delays, increases costs, and introduces risk as teams push code to production faster than it can be reliably validated.
Industry estimates place the global software testing and QA market at over $100 billion within the next few years. Yet much of that spend is still allocated to labor-intensive processes. Thunder Code’s proposition is straightforward. If even a portion of repetitive testing can be automated reliably, development teams can ship faster with fewer defects and lower overhead.
The company’s approach differs from traditional QA tools. Instead of requiring predefined test scripts, Thunder Code’s agents adapt to applications as they change. The system learns from feedback provided by developers and QA managers, refining test coverage over time. This adaptive model is designed to keep pace with modern continuous deployment environments rather than lag behind them.
Thunder Code operates as a subscription-based SaaS business. Pricing scales based on usage, typically linked to the number of applications or test agents deployed. The initial go-to-market strategy targets companies with in-house development teams, particularly those already struggling to balance release speed and quality. Early pilots and paying customers have been reported in France, Tunisia, the United States, and Canada, suggesting relevance across different engineering cultures and markets.
In 2024, Thunder Code raised a $9 million seed round. The round was led by Silicon Badia, with participation from Janngo Capital, XTX Ventures, and a group of well-known angel investors including Roxanne Varza and Karim Beguir of InstaDeep. For a company at an early product stage, the size of the round reflected confidence not only in the technology, but in the founders’ execution history.
That history matters. Jouini and Othmani spent nearly a decade building Expensya from an early-stage startup into a multinational SaaS company serving enterprise clients. They navigated long sales cycles, compliance requirements, and global expansion. Thunder Code benefits directly from that experience, as well as from the founders’ network of operators, investors, and early adopters.
The timing also works in their favor. Generative AI has shifted expectations across software development. Engineering teams are actively experimenting with automation beyond code generation, and testing is an obvious next frontier. Thunder Code enters the market at a moment when buyers are open to new workflows, even as they remain cautious about reliability.
Risks remain. The QA space is competitive, with established players such as BrowserStack, Tricentis, and Applitools investing heavily in AI capabilities of their own. Convincing teams to trust autonomous agents in production environments will take time. As with any AI system, maintaining accuracy and avoiding false positives will be critical to adoption and retention.
Still, Thunder Code’s early progress suggests momentum. The product exists. Customers are testing it. Capital is in place. Most importantly, the founders have already demonstrated that they can build enterprise software at scale and exit it successfully.
Karim Jouini and Jihed Othmani are not trying to reinvent software development. They are targeting one of its most persistent inefficiencies and applying a technology shift that engineers are already embracing elsewhere. If autonomous testing becomes standard practice rather than experimentation, Thunder Code is positioned to be part of that transition.
Click the ⤢ Icon to View the Magazine in full Screen for Best View
Designed as a seasonal publication, Voice of Development brings together research, reporting, and analysis meant to be read deliberately and revisited over time. Winter 2026 is a starting point: an attempt to answer, with clarity and restraint, what AIs can actually do—and what they cannot do.
Disclaimer: VoD Capsules are AI-generated. They synthesize publicly available evidence from reputable institutions (UN, World Bank, AfDB, OECD, academic work, and other such official data sources). Always consult the original reports and primary data for verification.
About the Author
