Today, Austin-based startup Driver officially emerged from stealth mode, unveiling an AI-powered platform designed to streamline the time-consuming process of onboarding engineers and creating technical documentation. The platform, which leverages advanced large language models (LLMs), promises to significantly reduce time-to-market for product development, particularly in industries such as semiconductors, where complex codebases and extensive documentation requirements are the norm.
Driver claims to help engineers craft tech docs and onboard twice as fast. Image (modified) used under Adobe Stock license
Alongside this launch, Driver announced an $8 million seed funding round, led by Google Ventures (GV), with participation from Y Combinator and other investors. All About Circuits spoke with Adam Tilton, CEO of Driver, to learn about the company and its vision for the future.
Unburdening Engineers of Documentation and Onboarding
Engineering teams, particularly in industries such as embedded systems and IoT, often face lengthy onboarding times when hiring new developers or migrating to new technologies.
“When we talk to customers, we hear it takes anywhere from three to six months for an engineer to onboard onto a new piece of technology. And this doesn’t just mean when they join the company,” Tilton explained. “Every single new project is an onboarding, and engineers constantly have to to learn how to use something new.”
Similarly, the creation and maintenance of technical documentation—such as user guides, reference manuals, and API documentation—are error-prone tasks that consume significant engineering time and require manual effort over several months.
Driver can help create installation instructions, feature overviews, integration guides, and more. Image used courtesy of Driver
Driver designed its platform to solve these problems by automating much of the technical documentation process and accelerating the time it takes for engineers to understand and work with complex codebases. The platform’s AI-driven approach enables organizations to quickly decode and document technology so engineers can focus on high-priority tasks rather than spend time parsing through extensive manuals and source code.
How Driver Works
Driver’s platform centers around large language models that automate the decoding and documentation of complex technologies. The tool syncs directly with codebases stored in version control systems such as GitHub or other source code management (SCM) tools. Once integrated, Driver can generate comprehensive technical documentation in a fraction of the time traditionally required, reducing the process from months to hours.
Nature language processing in deep learning-based models. Image used courtesy of Softermii
The platform employs a multi-layered architecture to guarantee efficiency and accuracy in its operations.
“When we consume a technical asset, we don’t start with a large language model. We do static analysis of it first,” Tilton said. “We consume it, we understand it, we break it down. We use static analysis tools to more deeply understand it, and then we work with large language models.”
The static analysis phase breaks the codebase into components, such as functions, classes, data structures, and control flows. This creates a foundational understanding of the code, which is then passed to the LLMs for natural language generation. The LLMs, trained on large corpora of technical content, generate detailed descriptions that are context-aware and tailored to the specific codebase being analyzed. This documentation can include functional descriptions, control flows, dependencies, and even high-level system architecture—all customized to the organization’s specific needs.
Using AI, Driver can quickly and clearly answer customer queries. Image used courtesy of Driver
Driver’s platform is also language-agnostic, supporting a wide range of programming languages and hardware description languages. This includes traditional languages like C, C++, Python, and Rust, as well as system languages like Verilog and VHDL.
In addition to supporting various languages, Driver’s platform includes a unified search function that allows engineers to search across assets, source code, and documentation. This feature allows users to navigate complex projects to find relevant sections of the codebase or documentation without manually combing through hundreds or thousands of files.
Just How Much Time Can Driver Save?
Driver built its platform to provide quantifiable time savings for organizations. According to Tilton, the platform can automate the creation of source code documentation in as little as two hours—a task that typically takes engineers up to three months to complete manually. Similarly, Driver’s platform reduces the time required for engineers to onboard to new projects by 50%, enabling teams to become productive in half the time.
From left to right, All About Circuits editor-in-chief Jeff Child, Driver AI CEO Adam Tilton, and Driver AI co-founder and CTO Daniel Hensley.
These performance improvements have a direct impact on product development timelines. By accelerating the onboarding process and automating technical documentation, Driver helps organizations bring products to market more quickly and efficiently. This is particularly important in industries like semiconductors, where time-to-market can significantly affect competitive positioning and profitability.
Designed to Help Engineers Do What They Do Best
Driver envisions a future where the complexity and delays associated with technical documentation and onboarding are no longer roadblocks for engineering teams. The platform aims to fundamentally reshape how organizations manage their technical assets by harnessing the power of AI and large language models. Ultimately, the company aims to make technology instantly understandable and allow engineers to focus on innovation rather than navigating extensive and outdated documentation.
