Jetbrains adds some IDE AI; Qdrant's GPU boost

Welcome to Runtime! Today on Product Saturday: Jetbrains has a take on the inevitable collision of AI and IDEs, Qdrant's vector search performance gets a GPU upgrade, and the quote of the week.

(Was this email forwarded to you? Sign up here to get Runtime each week.)

Ship it

A midsummer night's code: Generative AI has changed the way software developers think about the tools they employ to get the job done, and code editors and integrated development environments [IDEs] will never be the same. Jetbrains has carved out a niche with that community with its take on the IDE, and this week it jumped on the AI coding assistant train with the release of Junie.

Junie allows developers to "fully delegate routine tasks to your very own personal coding agent or collaborate with it to execute more complex ones together," the company said in a blog post. Jetbrains also said Junie scored very well on the SWEBench Verified suite of tests, but most developers won't get a chance to test those claims given that Junie is only available through the company's Early Access Program.

The plot thickens: Graph databases such as TigerEye's flagship cloud product are designed to establish relationships between different types of data. This week the company introduced a major update called Savanna, which the company said allows "storage and computation to scale independently and without size limits, to meet the changing workloads and growing data volumes required by AI adoption within enterprises."

TigerEye said the Savanna update is 6X more powerful than the previous version and 25% less expensive. "Proponents of graph databases argue that the native storage of relationships means that graph databases are intrinsically more efficient than relational databases for applications and use cases that depend on identifying and navigating the connections between entities," ISG Ventana Research's Matt Aslett told TechTarget.

Vector/Victoria: Speaking of databases, vector databases are a key component of generative AI applications that want to implement RAG (retrieval-augmented generation), which is one of the most widely used techniques for tamping down the hallucination problem. While traditional database companies think vector search is just a component they can add to their existing products, Qdrant is one of several startups nurturing a purpose-built vector database, and this week it released a new version that can tap into the power of GPUs.

Qdrant 1.13.0 "reduces indexing times, making it a game-changer for projects where speed truly matters," the company said. The new release supports Nvidia, AMD, and Intel GPUs, and "with the new feature the company can accelerate index building up to 10 times by using GPUs instead of central processing units," Qdrant co-founder and CEO Andre Zayarni told SiliconAngle.

Just one ping: Any business trying to sell a catalog of products relies heavily on its search capabilities, which needs to serve customers the right product in response to the right query. Perplexity has raised a ton of money to fuel its attempt to redefine search technology through generative AI, and this week it announced plans to allow third-party developers to incorporate its technology into their own apps through a new API.

Sonar is a new API service that allows customers to "build your own generative search capabilities powered with unparalleled real-time, web-wide research and the Perplexity features you've come to expect, like citations," Perplexity said in a blog post. Two versions are available: the basic Sonar APi and a Sonar Pro API, which "can handle in-depth, multi-step queries with added extensibility, like double the number of citations per search as Sonar on average," the company said.

Ringing twice: Speaking of APIs, Postman enjoyed substantial growth over the last decade as a centralized platform for building and sharing APIs. Any enterprise company looking for growth these days, however, has a giant sign labeled "AI" on their front door, and this week Postmand launched AI Agent Builder.

The new service allows "developers to quickly design, test, and deploy intelligent agents by combining LLMs, APIs, and workflows into a unified solution," the company said in a press release. Nearly every enterprise software company is working on a similar product, of course, but Postman co-founder and CEO Abhinav Asthana raised a good point in an interview with TechCrunch: “In my opinion, the cloud providers always prioritize consumption of the cloud over developer experience, and I think they’re willing to partner with anyone who does actually give the right developer experience."

Stat of the week

Anything that can improve the efficiency of data centers over the rest of the decade will be warmly embraced by the hyperscalers spending billions on massive buildings full of servers and storage, especially if it's simple. Researchers from the University of Waterloo said this week that they have identified a better way for Linux servers to process network traffic that could reduce energy consumption by up to 30% by tweaking just a few lines of code.

Quote of the week

"Look, all I know is, I’m good for my $80 billion.”  — Microsoft CEO Satya Nadella, commenting on OpenAI's audacious plans for the Project Stargate AI infrastructure buildout amid speculation that there's no way it has $500 billion in funding lined up. Microsoft has said it plans to invest $80 billion in capital expenditures — mostly to support its AI ambitions — during its 2025 fiscal year.

The Runtime roundup

Meta plans to invest as much as $65 billion in capital expenditures this year to support its own AI ambitions, CEO Mark Zuckerberg said Friday, reminding everyone that he's also a big boy.

Blackstone agreed to spend $1 billion on a natural-gas power plant in Northern Virginia, home to one of the largest groups of data centers in the U.S., according to Reuters.

Thanks for reading — see you Tuesday!