Participated · Series B · Mar 2025

MemryX is a fabless semiconductor company that designs edge AI processing solutions using a proprietary compute-at-memory technology and dataflow architecture. Its MX3 accelerator has reached production quality and the company says more than a dozen companies were competitively selected to enter production with MemryX. The MX3 is being designed into solutions such as AI-powered video management systems, industrial PCs, and edge servers. MemryX reports the MX3 delivers higher system performance while using less than 10% of the power of mainstream GPU solutions for vision-based AI inference. The company emphasizes a software-first core architecture and provides a public MemryX Developer Hub with tools and benchmarks across hundreds of models, which it credits with capital-efficient development and support for dozens of customers. MemryX plans to ramp MX3 production and complete the design of its next-generation MX4 accelerator and is headquartered in Ann Arbor, Michigan, with offices in Bangalore, Taipei, and Hsinchu, Taiwan.

Participated · Seed · Jun 2023

BlastPoint offers an AI-powered customer intelligence platform designed to help businesses better understand, reach, and serve their customers through auditable, hallucination-free predictive analytics that use only permission-granted data. Its technology is used across 13 verticals—most notably utilities and financial services—within a $13 billion addressable market. The platform supports programs that match financially struggling customers with assistance options such as income-capped payment plans and balance forgiveness. BlastPoint reports more than 2× year-over-year growth, 100 % customer retention, and coverage that touches nearly one-quarter of U.S. households. Future plans include building next-generation prediction technology while maintaining ethical AI standards and expanding solutions that improve equitable customer engagement. With a proven track record in energy and finance, BlastPoint positions itself as a trusted partner for organizations seeking responsible AI solutions.

Participated · Series A · Sep 2022

Voxel51 provides open-source and commercial software that enables teams to build datasets and computer vision models for machine learning and AI applications. Its open-source FiftyOne tool has been integrated into engineers' and scientists' workflows, while enterprise customers use FiftyOne Teams to securely collaborate on datasets and models. Customers span industries including automotive, robotics, security, retail and healthcare, and include LG Electronics, Berkshire Grey, Precision Planting, RIOS Intelligent Machines and Forsight. Led by CEO Brian Moore, the company sells both community and enterprise offerings to support development and deployment of computer vision systems. Voxel51 intends to use new funding to expand go-to-market operations and accelerate its product roadmap. Voxel51 develops FiftyOne, a platform and open-source tool for visualizing, analyzing and managing unstructured datasets used in computer vision and machine learning. FiftyOne launched open source in August 2020; this week Voxel51 introduced FiftyOneTeams, a fully managed edition with enterprise features like searching cloud-stored image and video datasets, dataset sharing, and version management. The company positions itself as a full‑stack data-infrastructure solution that avoids copying media or metadata onto its systems and aims to be a single source of truth for organizations’ visual data. Founders Jason Corso and machine learning PhD Brian Moore built the product to help ML engineers and data scientists verify and select data that will train models to perform desired tasks. Voxel51 says it will use new capital to further develop the platform and to double headcount from 13 to 26 employees by year-end. The company faces competitors such as Aquarium Learning and Lightly.AI but reports growth alongside rising demand for unstructured vision data. Corso declined to disclose revenue when asked. Voxel51 provides a video understanding platform for automotive, smart city and security applications. Its computer vision and deep learning models automatically identify and classify actions in video scenes and can comprehend activities and patterns. The scalable platform enables customers with large video datasets to search for specific objects, actions and behaviors and to automatically tag and integrate detected content into workflows and human decision-making processes. Founded in 2016 by University of Michigan professor Dr. Jason Corso and Dr. Brian Moore, the company is based in Ann Arbor, Mich. Voxel51 has completed a number of successful pilots in the automotive, smart city and security spaces. It closed a $2M seed round and recently concluded a $1.25M NIST grant to support development and will use the seed funds to launch its solution; total investment to date is $3.25M.

Participated · Seed · Oct 2021

Pulsora (formerly pulsESG) offers a centralized platform for enterprise sustainability management, covering data collection, disclosures, audit logging, analytics, and real-time target monitoring. The company emphasizes flexibility, auditability, integration, and a user-friendly interface to help organizations measure, report, and improve ESG performance. Pulsora serves over 400 corporations worldwide, including customers such as Apax Partners, Aker ASA, Aligned Climate Capital, Clayton Dubilier & Rice, and Cornerstone Building Brands. Founded in 2021 as a public benefit corporation, Pulsora has rebranded to reflect expanded platform capabilities. With the new funding, the company plans to bolster functionality—explicitly expanding carbon accounting—and prepare customers for forthcoming regulations. Management says funds will support global expansion across North America, Europe, the Middle East, Australia, and Asia within a roughly $12 billion nascent sustainability management market. pulsESG provides an integrated, comprehensive SaaS platform for ESG compliance tracking, disclosure, and analytics. The platform offers auditable deep analytics and can integrate with customers' source and destination ESG systems, giving end-users flexibility as regulatory requirements change. pulsESG plans to integrate with Workday Financial Management and Workday Human Capital Management to translate structured and unstructured data into financial information for disclosure management and external reporting. The integration aims to let customers assign ESG goals, track performance against those goals, and report metrics to regulators, investors, customers, and other stakeholders. Built-in analytics measure performance against goals and benchmarks and enable companies to improve practices based on insights. Founded in 2021 and headquartered in San Francisco, pulsESG is positioned to serve purpose-driven enterprises seeking comprehensive, auditable ESG solutions. pulsESG is a public benefit corporation focused on enabling purpose-driven enterprises to measure, manage and improve their ESG footprint. The company offers ESG management capabilities aimed at organizations prioritizing impact. pulsESG joined Accenture Ventures' Project Spotlight, an engagement and early investment program. Accenture (NYSE: ACN) made a strategic investment in the company through Accenture Ventures. The company had previously raised $10 million in seed funding from impact-driven investors. pulsESG is reported from San Francisco, CA. pulsESG is an integrated enterprise ESG SaaS platform that provides comprehensive, auditable, model-driven analytics and integrates with both internal data sources and external investor-facing systems. The product is designed to help enterprises track, analyze, derive insights for ESG compliance and improvement, and adapt to changing regulatory and stakeholder requirements. Key features highlighted include a model-driven architecture, full integration with source and destination systems, and end-user flexibility to modify models and mappings. The company emerged from stealth and expects to make the product generally available in Q1 2022. pulsESG is structured as a public benefit corporation, was founded in 2021, and is based in San Francisco. The company announced $8.5M in seed funding to support product launch and go-to-market activities, and is led by co-founders Murat Sönmez and Inderjeet Singh.

Participated · Seed · Mar 2021

Refraction AI develops the REV-1 last-mile delivery robot built on a bicycle foundation and designed to operate in bike lanes and roads rather than on sidewalks. The REV-1 can reach speeds up to 15 miles per hour and is currently being tested with a small fleet in Ann Arbor. The company uses cameras instead of LIDAR to lower costs and enable faster fleet scaling, and says the approach is resilient to Upper Midwest weather though it has limitations. Refraction was founded by University of Michigan professors Matthew Johnson-Roberson (now CTO) and Ram Vasudevan and debuted at TechCrunch Sessions: Mobility in 2019. The startup raised a $4.2M seed round to fund R&D, expand its geographic reach, and recruit grocery store and restaurant customers. With the new funding the company plans to expand operations beyond Ann Arbor, though no additional test markets have been announced. Refraction AI builds and operates the REV-1, a semi-autonomous, three-wheeled delivery robot used for restaurant deliveries and, more recently, grocery deliveries in Ann Arbor. The REV-1 is roughly the size of an electric bicycle, carries about six bags, travels 10–15 mph, and uses 12 cameras plus redundant radar and ultrasound sensors rather than lidar. The robots can navigate in rain and snow, do not depend on high-definition maps, and are being operated remotely with staff working from home. Refraction began grocery service from Produce Station after a three-month restaurant pilot and reports a 3–4x pandemic-related uptick in demand; as of May 2020 it had eight robots in operation and expected to have 20 soon. Operations have been adapted for COVID-19: robots are disinfected between deliveries, contain microbe-killing UV lights, and cleaning staff maintain physical distance. The company charges restaurants a flat $7.50 per delivery, has served over 500 customers (who may pay part of the fee if the business chooses), and tips go directly to its partners. Launched in July 2019, Refraction recently raised additional funding, bringing its total capital raised to $6.5M.