LibLab, which bills itself as an “SDK-as-a-service” platform for engineers, today announced that it raised $42 million in Series A funding from Insight Partners with participation from Zeev Ventures, Stepstone, Sheva, and Rainfall. Co-founder and CEO Sagiv Ofek said that the new capital will be put toward building out the company’s core service while expanding LibLab’s private beta.
Companies release software developer kits, or SDKs, to enable developers to use their API-based services. But creating an SDK can be arduous work. Each language and operating system has sets of requirements, and there’s the potential that security vulnerabilities and bugs crop up in the course of development.
One source estimated the cost of building an SDK in a single language at over $50,000. This led Ofek, an ex-Facebook and -Amazon engineer and the founder of dinner club app HomeDine (which TC previously covered), to investigate a way to automate the production and maintenance of SDKs.
Launched in stealth last year, LibLab provides SDK development tools that integrate with an API for authentication, error handling, security, and more. The platform generates a documentation page and developer portal, ostensibly ensuring that the API, SDK, and documentation remain in sync. Beyond this, LibLab monitors and updates the SDK “when the language evolves,” according to Ofek, and shows metrics that indicate how the API is being used.
“[LibLab is] code that writes code, allowing companies to write and support SDKs in multiple languages at once. Customers will not need to know multiple languages and will still be able to generate best-in-class SDKs to offer their developers,” Ofek told TC in an email interview. “LibLab saves companies time by generating SDKs in multiple languages [and] having best practices in place for each language. Companies with APIs can now offer SDKs to their developers, with proper documentation and security monitoring.”
Ofek offered scant details when pressed on the platform’s underpinnings, like whether LibLab uses AI to generate code and how the platform “knows” whether to update a given SDK. But he claims that LibLab is able to automatically generate nearly “everything” — referring to the SDK and documentation — all at once with “minimal effort from the customer.”
“LibLab’s code is very deterministic, parses API specs (e.g., from OpenAPI, Swagger, and Postman), and is based on pre-set rules to generate outputs in different languages,” Ofek said. “The core ‘code-gen’ technology will be open-sourced soon. We use TypeScript as our main code language to make it accessible to everyone, but each individual language (Java, Python, C#, etc.) has its own unique scaffolding template. The documentation is also parsed and generated the same way.”
No technology is perfect, and, barring evidence to the contrary, LibLab’s tool likely makes mistakes (assuming it even works as advertised). But it’s true that code-generating systems have become more capable in recent years with the advent of sophisticated machine learning techniques. For example, OpenAI’s Codex model can generate code in dozens of languages. Meanwhile, Facebook’s open source TransCoder tool can translate between C++, Java, and Python.
More should become clear when LibLab open sources its tools, which it plans to do in the coming weeks. Following the well-trodden path of open-source-developers-turned-SaaS providers, LibLab says it plans to make money by providing premium services on top of its open source offering
“It’s very hard to write code that writes code in multiple languages simultaneously. The main challenge is to ensure the API, SDK, and documentation are all in sync, all the time,” Ofek said. “One thing we’re doing to make LibLab as close as possible to perfect building in the open. Our core engine will be open source so developers can contribute and make our SDK generator better with time. Offering open source tools will enable the developer community to expand the languages LibLab supports.”
To date, LibLab has raised $50 million in venture capital.