Participated · Equity · Jun 2026

Neurometric builds an automated token engineering platform that evaluates individual model calls, modifies prompts, routes tasks to the most cost-effective models, and creates specialized small language models when needed. Its Task Endpoint Manager continuously updates model performance and pricing data to meet customer accuracy, cost, and latency requirements, and its Auto-SLM Creator generates purpose-built SLMs for recurring workloads. The platform also includes a marketplace of pre-trained task-specific SLMs for common workloads. Neurometric reports that routed or created models in early customer engagements have outperformed frontier models by as much as 20 accuracy points while reducing cost and latency. The company launched the platform publicly and closed $4 million in funding earlier this year. Neurometric plans to use the funding to expand its engineering and AI research teams and to add more optimization tools to the core platform.

Participated · Equity · May 2026

Patina develops novel scent molecules by modeling odor receptors and applying molecular design and machine learning, centered around a foundational model called Sense1. The team aims to create a universal code of smell and taste that can reconstruct rare natural ingredients and expand the palette available to perfumers and flavorists. Patina says its replications can simulate materials like rose oil at the biological level while consuming less water and petrochemicals. The startup is in talks with major fragrance houses and fashion brands about custom scents and is positioning itself against incumbents and peers such as Givaudan, Symrise, and startups like Osmo. Founded by Sean Raspet and Laura Sisson and launched last year, the company has moved into an office in Bushwick, Brooklyn and hired a small group of chemists. Its near-term plans include launching new molecules, funding research partnerships, and scaling computational simulations to accelerate molecule-to-market timelines.

Participated · Seed · May 2026

Dessn provides a design tool that runs customers' codebases in the cloud without setup by abstracting away runtime dependencies, enabling designers to work directly within a production environment. The product is intended for teams with an existing codebase who want to iterate on live projects rather than for ground-up ideation. Dessn says this approach eases handoffs to developers and supports sharing via public links. The company has early customers including Color, Wispr, and Mercury and offers a free tier (one repository compile and five prompts per week) plus paid plans starting at $39 per user per month. Dessn plans to add integrations such as Slack and meeting notetakers while explicitly avoiding a Figma integration. Founded by Gabriella Hachem and Nim Cheema about two years ago, the company currently has four employees and intends to remain small while hiring a few additional team members.

Participated · Seed · Apr 2026

Era builds a software orchestration layer that allows hardware makers to create AI agents and orchestrations for a variety of device form factors. The platform supports over 130 large language models from more than 14 providers to handle multimodal inputs and inference across devices like glasses, jewelry, and home speakers. Era was founded last year by CEO Liz Dorman, CTO Alex Ollman, and CPO Megan Gole. The company says its platform is designed to scale across millions of devices and to let users choose memory and model providers in a privacy-preserving way. Era does not manufacture devices itself but aims to enable third-party hardware makers and to work with the open-source and maker communities to showcase possible device experiences. Financially, the startup has raised $11 million to date across pre-seed and seed rounds.

Led · Seed · Feb 2026

Solaya is a French generative-AI startup that removes the cost and complexity of 3D capture by letting users create textured, rigged 3D models from simple smartphone videos. Its platform combines computer vision, diffusion models, and neural rendering, enabling brands to bypass expensive scanners and manual workflows. Early adopters such as LVMH, Mattel, Karl Lagerfeld, SNCF, and Dassault Systèmes already employ the technology for 3D commerce and robotics simulations, attracted by pricing that is five to ten times lower than traditional solutions. Led by co-founder and CEO Massimo Moretti, the team has experience building large-scale 3D pipelines and aims to make high-quality 3D capture as easy as taking a photo. With fresh capital, Solaya plans to expand its team, accelerate product development, and launch affordable self-service tiers for SMBs. Near-term goals include an app-store push, an 8× increase in user base by 2026, and the release of robotics SDKs. Longer term, the company aspires to be the dominant 3D data platform for “physical AI,” spanning warehouse automation to virtual try-ons in a market projected to reach $188 billion.