Participated · Seed · Aug 2026

Actualyze AI provides a hosted enterprise platform that routes every AI inference request through a governed path so organizations can apply access controls, inspections, billing, and audit trails before requests reach model providers. The platform integrates with OpenAI-compatible models, client tools, and third-party systems and is organized around four pillars: Govern, Secure, Operate, and Optimize. Actualyze ties each inference to the person, team, and application that made it, offers inline inference scanning and guardrails, and provides virtual model routing based on capability, cost, and quality with automatic failover. The company launched from stealth and is in Early Access via a Design Partner Program. Founders Rafi Khardalian and Sean Lynch previously founded Metacloud, acquired by Cisco, and the company announced a $7 million seed round backed by Storm Ventures, Canaan Partners, Morado Ventures, and AME Cloud Ventures.

Participated · Seed · Feb 2026

Quill is developing Quilliam, an AI agent designed to act as a sovereign Chief of AI Staff for modern professionals. The product begins as an automated meeting note-taker, then leverages the context gathered from users’ conversations to orchestrate tasks across writing, coding, research and collaboration tools. Its architecture keeps all user data on the device by default, offers optional end-to-end-encrypted cloud sync, and lets customers choose whether inference runs locally or in enterprise clouds such as Google Vertex or AWS Bedrock. This local-first approach appeals to security-conscious and highly regulated enterprises, as no user data is used for model training and the system can run fully air-gapped. Quilliam proactively executes workflows—creating tickets in Linear, updating Notion docs, or surfacing briefings—via its Model Context Protocol integrations. The company plans to use new capital to accelerate product development, expand customizable automations, and hire across growth and enterprise sales. Financially, Quill has just closed a $6.5 million seed round; no revenue or user metrics were disclosed.

Led · Seed · Feb 2026

General Legal launched in January 2026 to rebuild the law firm model around AI and human attorneys, aiming to automate repetitive legal work while preserving human legal judgment for strategy and negotiation. The firm signs clients via an engagement letter, embeds lawyers into customers’ workflows (including Slack), and offers features like a free legal health check to surface risks before fundraising or enterprise deals. The company reports reaching $1 million ARR within two months of launch. CEO Ryan Walker and co-founders Javed Qadrud-Din and J.P. Mohler all previously worked at Casetext; Walker helped build CoCounsel before Casetext was acquired for roughly $650 million. General Legal emphasizes transparent pricing and faster turnaround by using AI behind the scenes so attorneys can focus on higher-value legal work. The startup is using its YC backing and seed capital to scale its AI-native law firm model and expand access to proactive legal guidance for founders.

Led · Seed · Nov 2025

Code Four is a policing-technology startup founded by George Cheng and Dylan Nguyen that leverages artificial intelligence to turn body-camera video into preliminary incident reports officers can edit and submit in court or for record keeping. The platform also auto-generates transcriptions, interview summaries, and can redact both video and written materials for public-records requests. By automating these processes, the company aims to cut down on administrative workload and return officers to field duties more quickly. The product is sold on a subscription basis starting at $30 per officer per month, and the company says it is already serving 25 police departments. Code Four currently employs four people split between engineering and sales. The founders are alumni of Y Combinator and are slated to join the second cohort of the Palantir Startup Fellowship. While the system drafts reports with AI, final review and edits are performed by officers to address accuracy and ethical concerns.

Participated · Seed · Nov 2025

Hepta leverages a proprietary liquid-biopsy-native transformer model that evaluates up to one billion cfDNA fragments per blood sample, uncovering subtle epigenetic patterns linked to organ-specific disease. Its first application targets metabolic dysfunction-associated steatohepatitis (MASH), where the assay achieved a diagnostic AUC of 0.86 and cut false positives threefold versus standard FIB-4 blood tests. The company’s MASH Atlas dataset, built with Duke University and other hepatology centers, shows strong concordance between cfDNA methylation and liver-tissue biology, validating the approach as a true ‘liquid biopsy’ for the liver. Founded by former Illumina, Grail and Google leaders, Hepta aims to replace invasive biopsies with routine blood draws and provide longitudinal intelligence for chronic disease management. Near-term plans include expanding clinical datasets, scaling the platform, achieving regulatory clearance and forging pharma partnerships. The platform’s attention-based architecture positions it to move beyond liver disease to other chronic conditions. Hepta recently secured $6.7 million in seed funding to finance these milestones.