OpenAI Backs Isara, Google Launches Lyria 3 Pro

Several new developments in artificial intelligence demonstrate an increased pace in both investments and product launches. At the same time, multiple initiatives point to the need for better coordination, infrastructure, and control over AI systems.

OpenAI Invests in Isara for Coordination of AI Agents

The startup Isara has raised $94 million with support from OpenAI. The company is developing software that allows many AI agents to collaborate on complex tasks within the same system. The valuation is now $650 million, and the solution targets scenarios where a single model is insufficient.

The technology enables task distribution among multiple agents in one workflow. This provides better control over processes requiring parallel operations, such as data analysis and automated development.

Source: Wall Street Journal

Google Launches Lyria 3 Pro with Extended Music Generation

Google has launched Lyria 3 Pro, a new model for AI-generated music. Users can now create tracks up to three minutes long, an increase from the previous limit of 30 seconds. The model also offers more precise control over style and structure.

This expansion makes the tool more relevant for production rather than just experimentation. Musicians and content creators can use the model to produce complete tracks instead of short drafts.

Source: TechCrunch

GitHub Uses Copilot Data for Model Training

Starting April 24, GitHub will use interaction data from Copilot to improve its AI models. This includes code snippets and user input. Users will have the option to opt out of data collection.

The change allows better adaptation of models to real-world use since the training data reflects actual development tasks. At the same time, it increases the need for clear choices regarding privacy and data sharing.

Source: How-To Geek

Epic Microsystems Develops Solutions for AI Data Centers

Epic Microsystems has raised $21 million in Series A funding. The company develops technology for power and heat management in data centers running AI workloads.

The solution targets bottlenecks in modern data centers, where increased load demands more efficient energy use. Better control of temperature and power results in more stable systems and lower operating costs.

Source: Axios

ARC Launches New Benchmark for Real-Time Reasoning

The ARC Prize Foundation has introduced ARC-AGI-3, a benchmark that tests AI models in simple game-like environments. The focus is on real-time decisions rather than stored knowledge.

This provides a more practical measure of how models handle new situations. The method makes it easier to evaluate systems intended to operate in dynamic environments.

Source: Fast Company

Vultr Plans Expansion of AI Capacity

Cloud provider Vultr is seeking over $1 billion to build out AI data centers. The company previously raised $333 million in 2024 with a valuation of $3.5 billion.

The investment targets growing demand for GPU capacity. More companies need access to local infrastructure to run models without relying on a few large providers.

Source: The Information

What Does This Mean?

AIny brief assessment: Several of the news items point to the same trend: increased need for coordination of AI agents and greater demands on infrastructure. For Norwegian companies, this means increased focus on data centers, cost control, and integration of AI into existing systems. At the same time, access to computing power becomes an important competitive factor.

Read also: Global AI News: OpenAI and Kuaishou Updates