Participated · Seed · Jul 2026

Fish Audio develops AI speech generation and speech-to-text models and offers both open-source releases and paid API plans for creators, teams, and enterprises. The company has launched five models in the last year—three of which are open-sourced—and its S2.1 Pro model is available only through its paid API. Fish Audio reports more than 8 million users and $21 million in annual recurring revenue, and its GitHub Fish Speech repository has over 31,000 stars. It offers paid monthly plans that unlock generation minutes and voice cloning features, and it also runs enterprise APIs used by companies including HeyGen, Sanas and Plaud. The startup plans to release an audio understanding model this year and is building a speech-to-speech model as part of its roadmap. Fish Audio has also implemented an automated takedown process and creator verification flows to address consent and licensing concerns around uploaded voices.

Participated · Series B · Nov 2024

SuperAnnotate builds a low-code platform for creating, labeling, iterating on, and evaluating AI training datasets, initially focusing on labeling software and expanding into fine-tuning and model evaluation tools. The platform lets users connect local and cloud data, collaborate on projects, compare model performance by training data from a dashboard, and deploy models to various environments. It also operates a marketplace of crowdsourced annotators used for data-labeling tasks. The company counts clients such as Databricks and Canva and faces competitors including Scale AI, Weka, and Dataloop. SuperAnnotate was founded by brothers Vahan and Tigran Petrosyan and is San Francisco-based. The startup plans to use new funding to grow its ~100-person team, accelerate product R&D, and expand its roughly 100-company customer base while pursuing deeper customization for enterprise data fine-tuning. SuperAnnotate provides an AI-powered image annotation platform that supports manual labeling, pixel-accurate annotations, workflows, templates, dashboards, status tracking, and communication tools. The platform also helps teams run pilot projects with multiple annotation providers to choose the right partner. Led by CEO Tigran Petrosyan, the product targets data scientists and labeling teams with tools for recognition improvement and image status tracking. Launched this February, the company has already served more than 3,000 data scientists and over 100 companies. SuperAnnotate intends to use the new funding to continue expanding operations and broaden its business reach.