Participated · Seed · Apr 2026

Shapes builds a chat app where AI characters (called “Shapes”) participate alongside humans in shared group conversations, and those AI accounts are clearly labeled and can be created and customized by users. The company was founded in 2022 by Anushk Mittal and Noorie Dhingra and has grown largely via word of mouth to more than 400,000 monthly active users. Users have created roughly three million Shapes, and the app reports thousands of users spending two to four hours per day on the platform. Growth accelerated sixfold since the start of the year according to the company. Shapes positions itself as a next‑generation social chat product aimed at highly engaged online communities and fandoms. With the $8 million seed round, the company plans to accelerate development and fund user acquisition efforts.

Participated · Seed · Nov 2025

Poly builds an intelligent cloud file browser that leverages artificial intelligence to tag, summarize, and organize users’ documents while also enabling the generation of new content. Its core offering is a desktop application currently available for macOS, with a Windows release in the pipeline. The product aims to give individuals and teams a smarter way to store, browse, search, and share files from the cloud. Newly secured seed capital will be allocated to product development, expansion of AI infrastructure, and hiring for the early team. The company has opened a public waitlist alongside the launch of its desktop app, signaling the start of broader user onboarding. Poly is led by founder and CEO Abhay Agarwal, with Samuel Lo serving as Head of Design and Alexandre Seo as Head of Engineering. No operating metrics such as revenue or user numbers have yet been disclosed.

Participated · Seed · Jul 2025

Julius AI builds an AI data analyst that analyzes and visualizes extensive datasets and performs predictive modeling from natural-language prompts. It is designed to act like a data scientist, producing charts and answering queries such as visualizing revenue and net income correlations across industries and regions. The product was adapted for Harvard Business School’s new Data Science and AI for Leaders course after HBS professor Iavor Bojinov asked founder Rahul Sonwalkar to modify Julius for the class. Sonwalkar launched Julius after graduating from Y Combinator in 2022, having pivoted from a logistics startup he’d been building during the accelerator. The company reports more than two million users and generates over 10 million visualizations. Julius emphasizes focus on a specific data-science use case despite overlapping functionality with large language models like ChatGPT, Anthropic’s Claude, and Google’s Gemini.

Participated · Seed · Mar 2025

Momentic builds AI-driven tools that automate software testing and verification, allowing developers to describe critical user flows in natural language and receive fully generated tests. The platform targets pain points associated with traditional frameworks like Playwright and Selenium by emphasizing simplicity and speed. Momentic’s product is already used by 2,600 users at companies such as Notion, Xero, Bilt, Webflow and Retool, and it executed more than 200 million automated test steps in the past month. The startup recently added support for mobile environments and plans to expand into more sophisticated test-case management after hiring additional engineers. Co-founders Wei-Wei Wu and Jeff An draw on prior developer-tooling experience at Qualtrics and WeWork. While revenue figures were not disclosed, reported growth convinced investors to increase their backing. Momentic believes the proliferation of AI-generated code will drive demand for scalable, automated testing solutions that its platform aims to provide.

Participated · Seed · Dec 2024

Hamming.ai automates testing, monitoring, and governance for AI voice agents by deploying its own AI callers to place thousands of phone calls to client voice agents simultaneously. It provides LLM prompt management, automated red‑teaming to detect vulnerabilities, and call analytics to flag production issues and track user engagement. The company says its approach is 20× faster and 10× cheaper than manual testing and aims to further improve these margins through additional product development and refinement. Founded in 2024 and based in San Francisco, Hamming.ai targets reliability and compliance needs for industries governed by strict frameworks. The startup emphasizes that automated testing will be a cornerstone of trust‑focused AI strategies as regulations evolve. Financially, Hamming.ai announced a $3.8M seed round led by Mischief.