The Automated Code Testing Revolution is Underway: Announcing the SapientAI-8VC Partnership

May 16, 2023

AI and automation have become essential components in the developer's toolkit, transforming the way we build software and offering the best hope yet of meeting the unrelenting demand for scale, speed, and reliability that has come to characterize the cloud-native era. Whether as  engineers or investors, when attempting to identify the essential tools amid a sea of contenders, context is everything. This is doubly true for generating test code - a realization that helped to guide the formation of 8VC’s newest portfolio company, SapientAI.

The roots of today’s context-rich developer resources run deep. Autocompletion tools have been around since the 1960s, when Les Earnest pioneered programmatic spell checkers, while Alice Pascal included context-sensitive code completion in 1986. The advent of foundation model-powered developer tools, however, has finally ushered in a wholesale era of AI-assisted code authoring. Concurrently, the “shift left” phenomenon has concentrated more decision-making, as well as more responsibilities, with software developers. As testing and quality assurance have moved forward in the development process, bugs can be identified and eliminated sooner - but at substantial cost to developer time and attention.

Sapient offers a decisive answer to this conundrum, combining the power of foundation models with traditional lexical approaches to create the world's most comprehensive platform for developers to write and manage tests (including unit, functional, integration, performance, usability, and more). Foundation models need to consider combinatorial spaces spanning millions of choices. Lexical approaches can whittle these down to a thousand. This enables them to fit the context window of foundation models, which can then synthesize the optimal answer. Foundation models support Sapient’s mission of faster, more reliable software testing for everyone in three principal ways:

  1. Helping to generate precise test code of uncompromising quality.
  2. Automatically tying tests together across the stack, especially when the linkages are not obvious.
  3. Providing visibility into the health of tests across an enterprise, intelligently suggesting test updates based on changes in dependencies or performance, and modeling those updates for easy commits.

Historically, we have seen a number of attempts to build new testing platforms, but in addition to their limited utility and accuracy, they couldn’t deliver a strong user experience for all the different personae involved in software testing. Sapient’s platform, by contrast, addresses testing requirements across the entire stack: from frontend application developers, to backend services and data systems developers, to operational developers and SREs. Addressing each of these functions in a single platform is staggeringly difficult, yet it is an essential step toward maximizing one of the greatest virtues of foundation models: their ability to transfer learning between domains. For example, the model you train for frontend tests might be applicable to backend-services tests (and more). Sapient not only eliminates the manual test coding and guesswork that typically plague individual developers, but also maximizes the cross-stack utility of each test generated.

Rishi Singh, Sapient’s founder and CEO, previously co-founded Harness, the pioneering DevOps platform company. As CTO, he spent seven years solving the myriad challenges of continuous delivery, including the urgent need for accurate testing, through the thoughtful application of machine learning. Prior to Harness, Rishi led the development of Apple’s deployment automation platform, immersing himself in the universal problem of how to deliver rapid and significant changes without introducing new points of failure. His unique perspective, coupled with his ability to build core infrastructure, makes him the ideal leader to fulfill the promise of a next-generation, AI-powered software testing platform - and he has assembled a uniquely formidable team.

We believe that the future of AI-driven developer tooling lies in specialized, highly contextual platforms such as Sapient. By focusing on discrete stages of the development process, i.e. testing, feature creation, or refactoring, these AI developer tools will empower developers to write elegant code with unprecedented efficiency - doing what they do best while balancing all of the responsibilities their roles have come to embody. 

Today, Sapient emerged from stealth, backed by $5 million in seed funding led by 8VC. We are honored to partner with Rishi and friends as they set sail to reshape the landscape of AI-powered developer testing - and software development as a whole.

Continue Reading