TLDR: The Laude Institute, founded by Databricks and Perplexity co-founder Andy Konwinski, has announced the first recipients of its ‘Slingshots’ AI grants. This initiative provides rapid funding, computing resources, and engineering support to fifteen early-stage AI projects, primarily focused on advancing AI evaluation methods. Grantees are committed to delivering tangible outcomes like new companies or open-source projects, marking a significant step in fostering independent AI research.
The Laude Institute, a new initiative spearheaded by Databricks and Perplexity co-founder Andy Konwinski, officially unveiled the first cohort of its ‘Slingshots’ AI grants on November 6, 2025. This program, backed by Konwinski’s personal commitment of $100 million, aims to provide swift and effective resources for early-stage artificial intelligence research, thereby accelerating innovation beyond traditional academic confines.
The ‘Slingshots’ grants are designed as an accelerator for researchers, offering crucial support that is often difficult to access in conventional academic environments. This includes financial backing, access to high-performance computing resources, and hands-on assistance with product development and engineering. In return for this comprehensive support, grant recipients are expected to deliver a tangible outcome, which could manifest as a new startup, an open-source project, or another form of impactful deliverable.
The inaugural cohort comprises fifteen diverse projects, with a significant focus on tackling the complex challenge of AI evaluation. Among the selected initiatives are projects already gaining recognition, such as ‘Terminal Bench,’ a command-line coding benchmark developed in collaboration with Stanford researchers, which has seen rapid adoption, including by companies like Anthropic for its Claude 4 model. The latest iteration of the longstanding ‘ARC-AGI’ project is also part of this pioneering group.
New projects are also bringing fresh perspectives to established evaluation issues. ‘Formula Code,’ a collaborative effort from teams at CalTech and UT Austin, seeks to measure the proficiency of AI agents in improving existing codebases. Meanwhile, ‘BizBench,’ developed at Columbia University, introduces a comprehensive benchmark specifically for ‘white-collar AI agents.’ Further grants are dedicated to exploring innovative methods in reinforcement learning and model compression.
John Boda Yang, co-founder of SWE-Bench, is another notable participant, leading the new ‘CodeClash’ initiative. Drawing inspiration from SWE-Bench’s successes, ‘CodeClash’ will implement a dynamic, competition-driven model for code evaluation, a methodology Yang believes will significantly advance the field.
Also Read:
- AI2050 Scholarship Awards $18 Million to 28 Global Scholars, Advancing Human-Centered AI Research
- Inception Secures $50 Million Seed Funding to Pioneer Diffusion Models for Code and Text Generation
This ‘Slingshots’ program is part of the Laude Institute’s broader two-pronged research strategy, which also includes ‘Moonshots’ projects. While ‘Slingshots’ focuses on rapid impact from early-stage research, ‘Moonshots’ are dedicated to addressing long-term societal challenges in critical areas such as health, education, scientific discovery, and workforce reskilling. The institute’s approach underscores a commitment to fostering independent and beneficial research, bridging the gap between academic breakthroughs and real-world application.
