Kombai Raises $4.5M to Revolutionize UI Coding with AI

Kombai Raises $4.5M to Revolutionize UI Coding with AI

Kombai, a startup focused on enabling seamless conversion of UI designs to code for front-end developers, has recently emerged from stealth. The company has announced a successful $4.5 million seed funding round, led by prominent investors Stellaris Venture Partners and Foundation Capital. Embracing the power of AI, Kombai is poised to revolutionize the UI design to code process and empower developers worldwide.

With the rapid advancement of hardware, the evolution of software has highlighted an urgent requirement for exceptional user interfaces. Emerging startups and industry-leading tech firms alike are to create distinctive user interfaces. However, front-end developers often face the challenge of swiftly coding these intricate designs. To address this issue, Kombai will be releasing their AI model for public research this Wednesday.

Kombai, a startup based in Palo Alto and operating from Pune, India, was established in April 2022 by former executives of Mindtickle, Dipanjan Dey (serving as CEO) and Abhijit Bhole (CTO). With a focus on UI design interpretation and code generation, Kombai employs a combination of deep learning and heuristics models. The team devoted over 16 months to groundwork before launching their venture. Discover how Kombai’s innovative approach can revolutionize UI design and development.

“The model tries to develop an intuitive understanding of the design… and then it starts generating the code,” Dey said in an interview.

Kombai, named after the Indian dog breed in honor of the dog-loving co-founders, employs a systematic approach to generate UI code. During this process, Kombai effectively organizes logical elements, resulting in a cohesive div structure while minimizing hardcoded widths and margins in the CSS.

The code generated is then divided into distinct components, each thoughtfully named to enhance comprehension and encourage potential reuse. By identifying patterns, Kombai identifies loops and conditions, replacing static text with variables. Lastly, it utilizes publicly available multi-modal large language models (LLMs) to enhance specific segments of its automatically generated code, ultimately culminating in a polished and refined end product.

According to Dey, the AI model is responsible for the majority of the final output’s character count, with the LLMs accounting for less than 5%. Despite the complex steps involved in generating code using AI, Kombai’s model ensures speedy results that save valuable time for front-end developers. The generated code can be easily downloaded or directly copied into developers’ respective IDEs. Furthermore, developers have the flexibility to modify the code based on their requirements and seamlessly incorporate it into their existing codebases.

According to Dey, developers usually allocate 25-75% of their work time to crafting UI code, including CSS styles, HTML’s Document Object Model (DOM), and framework-specific boilerplate. As the importance of design has grown for businesses, front-end development has become more challenging, but there is currently a lack of standardization in front-end technologies, as highlighted by the co-founder in a conversation with ProWellTech.

Prior to unveiling its model to the public for research purposes, Kombai engaged in a collaboration with over 500 developers over the course of the last six months as part of its private research preview. The startup holds aspirations of becoming the “reverse Dall-E for UI design” and bringing back the enjoyment of front-end development for the five million front-end and 15 million fullstack developers worldwide, as shared by Dey with ProWellTech.

At present, Kombai mandates front-end developers to possess a Figma account in order to sign up or integrate Figma’s API token for fetching designs from the interface design application. In spite of this requirement, Dey mentioned the potential for seamless integration of the model with various other design tools, including Adobe XD.

“The only reason we are not going to XD is primarily because of our bandwidth,” he said.

The seed round attracted participation from a group of 20 angels, including undisclosed SaaS CEOs and CTOs, along with a handful of late-stage investors.

Alok Goyal, a partner at Stellaris Venture Partners, remarked in a prepared statement, “Within Kombai’s team, we uncovered a rare combination of technological and product expertise. They adopt a fundamentally new approach to address the problem. We are genuinely impressed by the significant progress the team has made in developing the product, as well as the positive feedback from developers.”

Currently, Kombai boasts a workforce of 13 individuals consisting of senior front-end engineers and deep-learning specialists. However, the company has plans to expand its team in the coming months by hiring several additional engineers.

“Over the past decade, we’ve seen the emergence of excellent design tools like Figma and Adobe XD. However, these tools fall short when it comes to generating meaningful code for developers. On the other hand, language model-based tools such as ChatGPT and Github Copilot offer valuable code suggestions based on textual inputs, but they aren’t suitable for UI development.

As a result, developers are left with the arduous task of manually translating every aspect of a UI design into code, which is both frustrating and time-consuming. This is a significant challenge that demands a solution. That’s why we’re excited to partner with the Kombai team in their mission to empower the world’s 20 million developers. Our goal is to enable them to focus more on tackling complex and fascinating problems, rather than dealing with mundane CSS,” expressed Ashu Garg, Partner at Foundation Capital.

Kombai intends to allocate a portion of its commencing inbound operations, as well as ongoing investments in research and development. The focus will be on enhancing the foundational models and fostering compatibility with various libraries and frameworks commonly employed by developer teams.