Beyond LLMs: A Look at Jev
As AI agents take on more complex workflows, a new question is emerging: do we really need a frontier LLM for every decision?
Jev, a new System One model released recently, answers that question by using structured inferences and answers with probabilities, in a blazing fast fashion. The model is all the rage in Silicon Valley.
Rain Capital’s General Partner, Dr. Chenxi Wang, sat down with Vecna AI’s founder, Aaron Brown, to talk through what Jev is and Vecna’s assessment of Jev and other models in the same category. Some of the high level conclusions:
While TypeSafe reports much larger gains for Jev in ideal conditions, independent tests suggest roughly 5x faster and 30 to 60x cheaper than frontier models for single call triage.
Jev is one of the models in a growing category that includes CLM from Stanford and NVIDIA, Laya, GLiNER2.5 Decide, and CUA-S1. The idea is not new, but its data and API design are new.
While many are intrigued by this category of System One models, more testing is needed to have confidence that they produce results that are consistently better than a random coin flip.
runZero Joins Dragos in $4.2B Accenture Deal
Rain Capital portfolio company runZero has officially joined Dragos, alongside NetRise, following the close of Accenture’s majority investment in Dragos. The combination brings together asset and exposure intelligence, software supply chain visibility, and operational technology security to build a more complete view across connected devices, networks, cloud, software, and critical infrastructure.
HD Moore and the full runZero team are joining Dragos, with HD stepping into the role of Chief Research Officer. Together with the NetRise team, they will help integrate the technologies into the broader Dragos platform while continuing to serve existing customers.
The transaction marks Rain Capital’s sixth portfolio exit in the past eight months. Congratulations to HD, the runZero team, and everyone involved in reaching this milestone!
Chenxi at CMU: AI & Cybersecurity
Dr. Chenxi Wang returned to Carnegie Mellon University for Startup Week, joining leaders from DataTribe, Accenture, Accel, and CMU for the Swartz Center for Entrepreneurship’s Lab to Market panel on how AI is reshaping cybersecurity.
The discussion explored how AI is accelerating existing threats, introducing new vulnerabilities, and challenging traditional approaches to defense. As offensive capabilities become faster and more autonomous, the panel also looked at autonomous defense as an increasingly important part of the security landscape.
Beyond the technology itself, the conversation highlighted the need for stronger collaboration across academia, industry, and government to develop and deploy AI securely - bringing together research, real-world deployment, and security expertise as the field evolves.
Coming Up: Rain at SF Tech Week
October 8 | 11:00 AM–2:00 PM | San Francisco
Rain Capital is hosting an SF Tech Week event with Acrew Capital and J.P. Morgan for a conversation on The Search for Durable Advantage in AI.
When everyone has access to the same models, what actually makes a company different? We’ll explore where lasting differentiation comes from, including proprietary data, workflow ownership, infrastructure, distribution, and trust. More importantly, which of these advantages can actually endure as the AI stack evolves?
Keynote / Fireside Chat
Anu Triveti — Head of AI, Oracle
Saša Zdjelar — Chief Security Officer & Head of Product Security, NVIDIA
Dr. Chenxi Wang (moderator) — Founder & General Partner, Rain Capital
Founders’ Panel
Snehal Antani — Co-Founder & CEO, Horizon3 ($250M Series E at a $2B+ valuation)
Keith Stewart — Founder & CEO, Humanix ($18M raised, Series A led by Acrew Capital)
Girish Chandrasekar — Co-Founder & VP, Labs, Straiker ($64M Series A; $85M total funding)
Aaron Brown — Founder & CEO, Vecna (multi-agent orchestration; former AWS AgentCore)
Stephanie Cherrin (moderator) — Head of Platform, Innovation at J.P. Morgan
Together, they’ll share perspectives on what it takes to build, differentiate, and scale enduring AI companies.
Podcast with Tinfoil: Inside Private AI Inference
Rain Capital portfolio company Tinfoil was featured on Risky Business Features, where co-founder Tanya Verma breaks down how private LLM inference works, from trusted execution environments and hardware attestation to GPU isolation and verifiable deployments.
The conversation explores how organizations can verify the exact code and model processing their data while keeping sensitive information inaccessible to both Tinfoil and its infrastructure providers and whether private inference could ultimately become a privacy baseline for AI, much like TLS did for the web.