Social Media

Light
Dark

The hot new thing: AI platforms that stop AI’s mistakes before production

The rise of AI-assisted code generation is becoming increasingly evident. According to Scott Guthrie, Microsoft’s executive vice president of Cloud and AI, as of March, approximately 40% of the code uploaded to GitHub Copilot, an AI developer tool, was both AI-generated and untouched by human hands.

This trend has given birth to startups aiming to prevent AI-augmented code from causing disruptions, and investors are taking a keen interest in these ventures.

Just recently, an Israeli startup called Digma secured $6 million in seed funding for its continuous feedback platform. This platform operates locally on developers’ machines, assisting them in analyzing their code, including code generated by generative AI, to identify potential issues. Similarly, a San Francisco-based testing platform named Kolena received $15 million in funding to develop tools for testing, benchmarking, and validating the performance of AI models.

Today, a relatively new startup from the Bay Area, Braintrust, is unveiling its own funding round of $3 million. Braintrust positions itself as an “operating system for engineers building AI software,” helping them avoid undesirable outcomes resulting from AI models. For instance, developers constructing customer support chatbots can utilize Braintrust’s technology to ensure accurate responses, preventing the generation of incorrect information.

Braintrust boasts an impressive roster of backers, including renowned angel investor Elad Gil, who played a role in incubating Braintrust’s initial product. Other notable investors include Adam D’Angelo of Quora, Clem Delangue of the AI-focused HuggingFace, and Greg Brockman, co-founder of OpenAI.

While having a prestigious group of investors can be advantageous, whether Braintrust can lead the field remains an open question. Nevertheless, CEO Ankur Goyal is dedicated to addressing the challenge of ensuring AI code does not disrupt a company’s workflow.

Ankur Goyal’s journey is rooted in a passion for technology. Raised in Pittsburgh, he initially aspired to be a doctor but was captivated by technology, particularly during a high school linear algebra class that introduced him to Google’s PageRank algorithm. Goyal shifted his focus to computer science at Carnegie Mellon University, dropping out in his junior year to work on a relational database system at MemSQL, an early Y Combinator startup. After co-founding Impira and later becoming the head of its machine learning platform following its acquisition by Figma, Goyal recognized the growing challenge of building high-quality AI-driven software.

This realization prompted him to establish Braintrust in the summer, with the aim of addressing the unique difficulties posed by AI software development. Unlike traditional software, AI is inherently non-deterministic, making it challenging to guarantee the performance of code. Braintrust is collaborating with companies that have abundant data resources, such as Zapier and Coda, to beta test its product. Braintrust allows these companies to use their extensive datasets to evaluate their software effectively.

Furthermore, Braintrust operates within its clients’ cloud environments, overcoming compliance issues and facilitating adoption within enterprises.

Although Braintrust is in its early stages and competition in this field is intensifying, Goyal believes that his startup fills a void that existed in the AI development landscape. While continuous integration has evolved significantly over the past decade, a similar methodology and workflow for AI had not existed until Braintrust’s emergence.

Leave a Reply

Your email address will not be published. Required fields are marked *