Participated · Series C · Nov 2025

Founded in 2017, CoLab Software builds EngineeringOS, a cloud workspace that lets engineers run design reviews, annotate 2D/3D files, and apply AI agents to improve product quality. Its first agent, AutoReview, launched in June and already has more than 47,000 engineers on its waitlist, while early customers such as Ford, Lockheed Martin, GE Appliances, and Schneider Electric are using the core platform today. Management says these capabilities enable customers to shrink design cycles from months to hours and are driving seven-figure annual contracts. The company reports it is on track to nearly triple revenue in 2025, reflecting strong demand for AI in engineering workflows. CoLab plans to roll out additional AI agents, deepen integrations with other engineering software, and announce several strategic partnerships over the next 12 months. The platform also addresses a looming retirement-driven knowledge gap by capturing expert rationale that can be reused by future engineers.

Participated · Series A · Aug 2025

Arintra offers a GenAI-native platform that combines large language models with clinical knowledge graphs to interpret patient charts in context and apply specialty-specific coding guidelines. Every code the platform generates is fully explainable and auditable to build trust in high-stakes healthcare settings. The product integrates natively with Electronic Health Record systems including Epic and Athena and is designed to operate almost invisibly within existing clinical workflows. Leadership includes CEO Nitesh Shroff and CTO Preeti Bhargava. The company intends to use recent funding to accelerate product development and scale its team. Arintra also plans to open a new Bay Area headquarters to meet rising demand. Arintra builds a proprietary AI-driven platform that fully automates the majority of medical coding generation with no human intervention, using models built on top of open-source large language models. The product integrates without changing existing workflows for patients, clinicians, coders, or payers, enabling seamless deployment into hospital billing processes. Arintra’s models are trained on proprietary datasets and are designed to reduce claim denials, under-coding, and long receivable cycles, materially improving hospital revenue and cash flow. The founders, Nitesh and Preeti, hold PhDs (computer vision and computer science) from the University of Maryland and developed the platform over the past two years. The company targets a large, vertical market: US medical coding inefficiencies that the article estimates cost hospitals over $100 billion annually amid ~5 billion claims per year. Arintra builds an AI-powered assistant that automates clinical documentation and billing to help doctors save time and increase revenue. The product targets physician workflow by reducing administrative burden through automated documentation and billing processes. The team combines medical experts and AI specialists to develop the assistant. Arintra reports partnerships with 30+ healthcare organizations. The company states a mission to create the world’s most intelligent assistant for doctors and is positioning to scale that capability. It recently announced accelerator backing from Y Combinator.

Participated · Series A · Jul 2025

Castellum.AI is a financial-crime compliance platform that combines in-house risk data, AML/KYC screening, and explainable AI agents in a single, audit-ready system. Its platform collects sanctions, PEPs, adverse media and other risk data from original sources and uses machine-learning enrichments and AI summaries to accelerate reviews and monitoring. The company reports reducing AML/KYC false positives by 94% and time spent on compliance reviews by 83% out of the box, and cites client results across sponsor banks, community banks, a Fortune 50 firm and crypto exchanges. Castellum.AI’s AI agents passed a CAMS practice exam on the first try, and the company is seeking an official proctored CAMS certification from ACAMS. It plans to deepen integrations with financial institutions through modular implementations that layer AI agents on top of existing tech stacks to reduce switching costs and speed deployments. Detailed audit trails and explainability are emphasized to align deployments with regulator expectations and provide oversight for front-line financial-crime prevention. Castellum.AI offers a global risk data platform for compliance teams, delivering sanctions, PEPs, adverse media and other financial-crime datasets via an online platform, API, or bulk data subscription. Its patent-pending automated data collection, cleaning, standardization and enrichment process updates lists every five minutes and claims to reduce false positives by 80%. Since launching its beta in March 2022 the company has achieved over 1,250% user growth and signed Fortune 500 customers including major banks and insurers as well as global government clients. Over 100 news organizations and NGOs have also used its data for reporting and analysis. The product supports KYC onboarding and rescreening, transaction screening, crypto sanctions screening, supply chain compliance and related use cases. The company plans to use the recent funding to scale product and engineering resources and expand outreach to enterprise customers.

Participated · Series A · Feb 2024

Founded in 2022 by ex-Pivotal CEO Rob Mee, Mechanical Orchard uses AI-enhanced tooling and cloud instances to create modernized copies of companies' aging apps and services. The company leverages generative AI as a "pair programmer" while keeping developers in the loop to debug and review output. Its process maps legacy system behavior and dependencies, then writes cloud-hosted code that replicates the old system’s functionality; customers own the resulting code and can deploy it where they choose. Mechanical Orchard positions its approach around risk reduction and continuous proof of migration to build IT leader confidence. It says it has enterprise customers in retail and logistics in the pipeline and sees demand across manufacturing, transport and financial services. The San Francisco–based company has about 90 employees with offices in the U.K., Ireland, Italy and Germany. New funding will be used to expand the company’s AI capabilities through R&D. Mechanical Orchard offers a platform that moves large enterprises off legacy mainframe systems onto cloud-based applications using iterative, AI-enhanced and reverse-engineering approaches. The technology gives CIOs and CTOs enhanced visibility and control to reduce migration risk while keeping critical systems online. Successful migrations produce faster speed to market, improved R&D capabilities, and easier integration of new revenue models. The company says recent funding will support continued investment in platform research and development and a comprehensive global sales and marketing program, and it expects further hires across the business in the next year. Launched in 2022 and headquartered in San Francisco, Mechanical Orchard now totals 50+ employees and has an international presence in the UK, Ireland, Italy, and Germany. Leadership includes founder and CEO Rob Mee, co-founder and COO Matthew Work, and Chief Scientist Kent Beck.

Participated · Seed · Dec 2023

Velt offers a JavaScript SDK that lets developers add collaborative features — commenting, live walkthroughs, voice/video recording, huddles and AI-powered tooling — to any app in under 30 minutes. The product set also includes Superflow, a companion aimed at marketing agencies for collaborative work on websites, videos and PDFs. Several features use AI: automatic transcripts and summaries for recordings, comment categorization, and a contextual copilot for human-to-AI collaboration and in-context content edits. Velt targets creative tools, business analytics and CRM/task-management verticals and lists customers such as eqtble, awesomic, zoomforth and parc. The company declined to disclose exact customer count but is approaching three digits and says it doubled revenue in the past two months; pricing starts at $999 per month for up to 5,000 monthly active users and some large customers with >1M monthly users require custom pricing. Velt launched as part of Y Combinator’s Winter ’22 batch and was founded by former Google PM Rakesh Goyal. The 10-person team is distributed across the U.S., U.K. and India, and the firm plans to heavily prioritize AI on its roadmap.