Our AI writing assistant, WriteUp, can assist you in easily writing any text. Click here to experience its capabilities.

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

View Original View Raw

Summary

AI-assisted code is becoming increasingly popular, leading to a rise in startups that are creating tools to prevent AI mistakes before production. One such startup is Braintrust, which recently announced a $3 million funding round. Braintrust is an "operating system for engineers building AI software" which helps them to ensure their AI models are accurate. It is backed by renowned angel investor Elad Gil and other notable investors such as Adam D'Angelo of Quora and Greg Brockman of OpenAI. The startup is currently being beta tested by companies like Zapier and Coda and works by helping companies curate representative datasets to test their software. Deepchecks is another similar startup that recently raised seed funding.

Q&As

What percentage of code that developers are uploading to AI developer tool GitHub Copilot is AI-generated?
40%

What is the purpose of Digma, Kolena, and Braintrust?
The purpose of Digma, Kolena, and Braintrust is to keep AI-augmented code from mucking up the works and to build tools to test, benchmark and validate the performance of AI models, and to help developers avoid bad results from AI models, respectively.

Who are some of Braintrust's investors?
Renowned angel investor Elad Gil, Adam D’Angelo of Quora, Clem Delangue of HuggingFace, and OpenAI co-founder Greg Brockman are some of Braintrust's investors.

What problem does Braintrust aim to solve?
Braintrust aims to solve the problem of ensuring that AI code doesn't break a company's workflow.

How do Braintrust's customers use their product?
Braintrust's customers use their product to ensure that their chatbot answers questions accurately rather than hallucinating false information, and to help them curate "golden datasets" that they can use to test their software.

AI Comments

👍 This article provides an interesting insight into the hot new trend of AI-assisted code and the startups that are helping to ensure that AI code doesn't break a company's workflow. It's great to see renowned angel investors, like Elad Gil, taking notice of the innovative technology and backing it with their funds.

👎 This article does not provide much detail about the actual technology or the startups that are at the forefront of the AI-assisted code trend. It also fails to address any potential issues that may arise from the use of AI-assisted code in production.

AI Discussion

Me: It's about AI platforms that help prevent mistakes in production when AI-generated code is used. The article talks about a few startups that are working on this, and how investors are taking notice.

Friend: Interesting. What do you think the implications of this are?

Me: Well, I think it's a positive sign for the development of AI-generated code. These platforms will help developers build more reliable AI software, which will in turn help reduce the number of mistakes made in production. This could also help address some of the issues with data privacy and compliance that can arise when using AI-generated code. Overall, I think these platforms will help foster greater trust in AI-generated code and pave the way for further development in this area.

Action items

Research other AI-related startups that are developing tools to test, benchmark and validate the performance of AI models.
Explore the potential of using Braintrust’s technology to ensure that customer support chatbots answer questions accurately.
Consider the implications of using Braintrust’s product to curate “golden datasets” that can be used to test software.

Technical terms

AI-assisted: Refers to code that is generated with the help of artificial intelligence.
GitHub Copilot: A developer tool created by Microsoft that allows developers to upload code.
Digma: An Israel-based startup that provides a continuous feedback platform for developers to analyze their code.
Kolena: A San Francisco-based testing platform that builds tools to test, benchmark, and validate the performance of AI models.
Braintrust: A Bay Area-startup that provides an “operating system for engineers building AI software” to help them avoid bad results from AI models.
Elad Gil: A renowned angel investor who helped incubate Braintrust’s initial product.
Quora: A question-and-answer website.
HuggingFace: A buzzy AI outfit.
OpenAI: A research laboratory focused on artificial intelligence.
Continuous Integration: A software development practice in which developers regularly merge their code changes into a shared repository.
Deepchecks: An Israeli startup that provides “continuous validation for AI.”