Participated · Seed · Jul 2026

Cast Insights converts spoken audio from television, radio, podcasts and livestreams into a permanent, queryable intelligence layer that tracks who is speaking, what is said, when it occurs, and how narratives and sentiment evolve. The platform emphasizes real-time features such as alerts within thirty seconds of broadcast, speaker-level voice tracking, and narrative curves across thousands of sources. Founded in 2025 and based in San Francisco, the company has gathered more than 2.3 million hours of speech data from over 20 countries and 10 languages. Cast Insights plans to scale to 20 million hours processed and 500,000 hours of live content monitored daily to build a large American ephemeral speech dataset. The product is offered to institutional clients including hedge funds, policy teams, corporate strategy groups, and newsrooms. The company is backed by investors including Abstract Ventures, Village Global, HF0, Max Ventures, and Founder Embassy.

Participated · Seed · Jul 2026

MicroAGI develops Atlas, a platform that records demonstration data on factory lines to train control policies in simulation and then transfer those policies onto appropriate robots. The stack is designed to be hardware- and model-agnostic, allowing policies to run on different robot makers and model providers while streaming production data back to improve the fleet. The company targets large industrial manufacturers and positions deployment — not the robot hardware — as the primary constraint. MicroAGI has rapidly expanded its team and R&D footprint, notably growing its Greater Zurich Area research center from 10 to 100 researchers and hiring across Munich, Zurich, Istanbul and London. Founded roughly ten months before the announcement, the company closed a $55M seed round, the largest seed in German history, to fund scaling and deployments.

Led · Seed · Jun 2026

Sazabi builds an AI-native observability platform centered on chat, autonomous agents, and a logs-first approach that treats logs as the single source of truth for understanding production systems. The company aims to replace complex telemetry stacks and traditional dashboards by using AI to reconstruct views engineers need from log data and to proactively detect, investigate, and resolve issues. In a closed alpha it onboarded 50 teams in two weeks, ran 8,000 background investigations, detected 2,000 issues, and opened 200 pull requests against customer repositories. Sazabi says customers can onboard quickly — one cited customer reported a 15-minute onboarding and immediate useful alerts — and the product is being adopted by AI-native engineering teams. The company was founded in 2025 and is based in San Francisco. Following the seed raise, Sazabi plans to expand its engineering team, accelerate product development, and deepen integrations across cloud and developer platforms.

Led · Seed · Jun 2026

MNX builds a decentralized exchange offering futures, perpetuals and prediction markets tied to AI-related assets, including valuation futures, equity perpetuals focused on AI companies, and compute perpetuals using indices from SemiAnalysis. The platform uses 200-millisecond batch auctions enabled by MegaETH’s sub-millisecond block times to reduce latency arms races and MEV, and plans to offer leverage where appropriate. MNX targets trading and hedging across the AI value chain—compute, electricity, benchmarks, commodities, public equities and private lab valuations—and aims to attract tech workers, retail traders and crypto users using tokenized real-world assets. The company launched in the prior fall and is co-founded by Stephen Grugett and Ian Philips, with Philips stepping away from Manifold Markets; MNX is currently in private beta and intends to launch on mainnet later this summer. The team today consists of the co-founders and several part-time non-technical contractors and plans to hire three to six full-time employees to scale operations.

Participated · Seed · May 2026

General Compute operates an inference-focused neocloud that rents AI processing power for models during the active inference phase. The company has partnered with SambaNova and has $300 million of the vendor’s SN50 inference chips on order to power its offering. Its hardware choice emphasizes air-cooled, lower-power chips that can be colocated in existing data centers and repurposed crypto-mining facilities. General Compute launched its cloud offering recently and claims to be the fastest at running the MiniMax 2.7 open-source LLM. The founders, CEO Finn Puklowski and CTO Jason Goodison, are pursuing colocation deals to scale capacity without heavy infrastructure buildouts. Financially, the company completed a $15 million seed round at a $60 million post-money valuation led by FUSE VC, with participation from Carya Venture Partners, Village Global Ventures, and investment from Evercrest Capital Partners.