--- url: 'https://qubit.capital/blog/us-seed-weekly-funding-roundup-mar-13-20-2026' title: 'US Seed Weekly Funding Roundup (Mar 13-20, 2026): $84.0M Raised Across 4 Deals' author: name: Sahil Agrawal url: 'https://qubit.capital/blog/author/sahil' date: '2026-03-20T03:41:30+05:30' modified: '2026-03-20T15:19:14+05:30' type: post summary: 'US seed-stage startups raised $84M this week across cybersecurity, deeptech, and AI infrastructure. RunSybil led with $40M for autonomous penetration testing.' categories: - Weekly Funding Roundup image: 'https://qubit.capital/wp-content/uploads/2026/03/featured-us-seed-64372.webp' published: true --- # US Seed Weekly Funding Roundup (Mar 13-20, 2026): $84.0M Raised Across 4 Deals US seed-stage startups raised $84 million this week across four deals, with cybersecurity and AI infrastructure commanding the lion’s share of investor attention. RunSybil’s $40 million round — one of the largest seed raises in cybersecurity this year — anchored the week, followed by Claros’s $30 million bet on reimagining data center power from the chip up. A clear theme runs through this week’s deals: the AI agent explosion is creating entirely new infrastructure and security categories. Two companies are building protection layers for the agentic era, one is solving the physical power bottleneck AI workloads create, and another is providing the email plumbing AI agents need to interact with the world. Investors are placing big bets that the picks-and-shovels of the AI agent economy are being defined right now. Weekly Funding Roundup MAR 13-20, 2026 $84M TOTAL RAISED 4DEALS CLOSED 100%SEED $21MAVG DEAL SIZE USTOP REGION BY STAGE Seed$84M100% BY SECTOR RunSybilCybersecurity$40M ClarosDeep Tech$30M ManifoldCybersecurity$8M AgentMailAI Infrastructure$6M Table of Contents - [1. RunSybil Raises $40M for AI Cybersecurity Testing](#1-runsybil-raises-$40m-for-ai-cybersecurity-testing) [Deal Overview](#deal-overview) - [Investor Profile](#investor-profile) - [Company and Leadership](#company-and-leadership) - [Problem and Opportunity](#problem-and-opportunity) - [Product and Technology](#product-and-technology) - [Use of Proceeds and Vision](#use-of-proceeds-and-vision) - [Market Context](#market-context) - [2. Claros Closes $30M for Chip-Level Power Infrastructure](#2-claros-closes-$30m-for-chip-level-power-infrastructure) [Deal Overview](#deal-overview-1) - [Investor Profile](#investor-profile-1) - [Company and Leadership](#company-and-leadership-1) - [Problem and Opportunity](#problem-and-opportunity-1) - [Product and Technology](#product-and-technology-1) - [Use of Proceeds and Vision](#use-of-proceeds-and-vision-1) - [Market Context](#market-context-1) - [3. Manifold Raises $8M for AI Detection and Response](#3-manifold-raises-$8m-for-ai-detection-and-response) [Deal Overview](#deal-overview-2) - [Investor Profile](#investor-profile-2) - [Company and Leadership](#company-and-leadership-2) - [Problem and Opportunity](#problem-and-opportunity-2) - [Product and Technology](#product-and-technology-2) - [Use of Proceeds and Vision](#use-of-proceeds-and-vision-2) - [Market Context](#market-context-2) - [4. AgentMail Raises $6M for AI Email Infrastructure](#4-agentmail-raises-$6m-for-ai-email-infrastructure) [Deal Overview](#deal-overview-3) - [Investor Profile](#investor-profile-3) - [Company and Leadership](#company-and-leadership-3) - [Problem and Opportunity](#problem-and-opportunity-3) - [Product and Technology](#product-and-technology-3) - [Use of Proceeds and Vision](#use-of-proceeds-and-vision-3) - [Market Context](#market-context-3) - [Lessons for Founders](#lessons-for-founders) ## 1. RunSybil Raises $40M for AI Cybersecurity Testing ### Deal Overview - **Stage:** Seed - **Sector:** Cybersecurity - **Geography:** United States - **Round Size:** $40 million ### Investor Profile Khosla Ventures led the round alongside S32, Anthology Fund (a joint vehicle from Anthropic and Menlo Ventures), Conviction, and Elad Gil. Angel investors include Nikesh Arora, Jeff Dean, Amit Agarwal, and leaders from OpenAI, Palo Alto Networks, Stripe, and Google. The participation of Anthropic’s investment arm is notable — it signals that leading AI labs see autonomous security testing as a critical counterweight to AI-driven threats. ### Company and Leadership [RunSybil](https://runsybil.com) was founded in 2023 by Ari Herbert-Voss, OpenAI’s first security research hire (joined 2019), and Vlad Ionescu, former offensive security red team lead at Meta. Their combined experience in AI research and adversarial security gives RunSybil rare credibility in a space where domain expertise is non-negotiable. ### Problem and Opportunity Traditional penetration testing is periodic, manual, and expensive. Security teams schedule discrete assessments managed by external firms, leaving gaps between tests where new vulnerabilities can emerge. As AI agents proliferate across enterprise operations — from procurement to legal to engineering — the attack surface is expanding faster than manual testing can keep up. ### Product and Technology RunSybil’s AI agent, Sybil, performs continuous autonomous penetration testing against live applications. It finds, exploits, and documents real security vulnerabilities without human intervention. The agent learns from real-world exploitation, building a compounding knowledge base. It is already deployed at Cursor, Notion, Turbopuffer, Baseten, Thinking Machines Lab, major financial institutions, and Fortune 500 companies. ### Use of Proceeds and Vision Funds will go toward scaling the platform, expanding the enterprise customer base, and advancing AI agent capabilities for vulnerability discovery across increasingly complex AI-powered environments. RunSybil aims to make security testing continuous and autonomous rather than periodic and manual. ### Market Context The penetration testing market is being reshaped by AI. Competitors include XBOW, which recently raised $120 million and achieved unicorn status, alongside traditional pentest firms. RunSybil’s AI-native approach and founding team pedigree position it as a serious contender in this rapidly consolidating space. ## 2. Claros Closes $30M for Chip-Level Power Infrastructure ### Deal Overview - **Stage:** Seed - **Sector:** Deep Tech - **Geography:** United States - **Round Size:** $30 million ### Investor Profile General Catalyst led the round with participation from Red Cell Partners, Systemiq Capital, Aero X Ventures, Trenches Capital, and VIPC. General Catalyst’s involvement is significant — the firm has been aggressively backing AI infrastructure plays, and Claros fits squarely into their thesis that the physical layer of AI needs as much innovation as the software layer. ### Company and Leadership [Claros](https://claros.com) is led by CEO Daniel Kultran, with operations spanning Torrance, California and McLean, Virginia. The company emerged from stealth in early 2025 with $9.75 million in initial funding and has already fabricated three integrated voltage regulator (IVR) designs currently being tested in their Los Angeles-area lab. ### Problem and Opportunity Data center power consumption is surging due to AI workloads, and current power delivery systems waste significant energy through AC-to-DC conversion and voltage regulation at every stage. This is a hardware-level bottleneck that software optimization alone cannot solve. As the global data center market is projected to exceed $500 billion by 2030, even small efficiency gains translate to massive cost savings. ### Product and Technology Claros builds a chip-to-grid power management platform combining two innovations. First, an integrated voltage regulator that delivers power directly to processing units, eliminating conversion losses. Second, a DC-native Power Gateway that reduces AC-to-DC conversion losses and integrates with multiple power sources. Together, they minimize energy waste at every level of the data center power chain. ### Use of Proceeds and Vision The funding will go toward expanding lab facilities, growing the team, continuing IVR and Power Gateway prototyping, and beginning first manufacturing runs. Claros aims to redefine data center energy from the chip all the way to the grid. ### Market Context Data center power is one of the most acute constraints on AI scaling. While most startups attack this problem through software-level optimization or cooling, Claros targets the fundamental physics of power delivery. Their hardware-level moat through proprietary chip designs creates a defensibility that is difficult to replicate. ## 3. Manifold Raises $8M for AI Detection and Response ### Deal Overview - **Stage:** Seed - **Sector:** Cybersecurity - **Geography:** United States - **Round Size:** $8 million ### Investor Profile Costanoa Ventures led the round with Cherry Ventures, Rain Capital, and Modern Technical Fund joining. Notable angel investors include former Uber CSO Joe Sullivan and former Google DeepMind CISO Vijay Bolina. The security executive angels bring both credibility and direct enterprise buyer relationships. ### Company and Leadership [Manifold](https://manifold.security) was founded by Neal Swaelens, Oleksandr Yaremchuk, and Michael McKenna. Swaelens and Yaremchuk previously co-founded Laiyer AI, where they created LLM Guard — the most widely adopted open-source LLM firewall. That open-source track record gives them deep expertise in AI security patterns and instant community credibility. ### Problem and Opportunity As autonomous AI agents gain access to enterprise systems, databases, and tools, the attack surface expands dramatically. Security teams have no visibility into what agents are actually doing — which tools they call, which systems they access, and what actions they take. Existing security tools were built for human users, not autonomous agents. ### Product and Technology Manifold’s AI Detection and Response (AIDR) platform provides runtime security for autonomous AI agents on enterprise endpoints. Security teams get a real-time map of every agent, their connections to MCP servers, databases, and external systems, with anomalies flagged as behavior drifts. The platform deploys agentlessly, leveraging existing infrastructure with no new architecture, gateways, or proxies required. ### Use of Proceeds and Vision Funds will fuel development of the AIDR platform to protect companies from the expanding risks of autonomous AI usage. Manifold is positioning itself as the essential runtime security layer for the agentic AI era. ### Market Context AI agent security is an emerging category. Competitors include Prompt Security, Lakera, and Protect AI, but most focus on prompt-level protection. Manifold’s differentiation lies in runtime agent behavior monitoring — watching what agents do, not just what users ask them to do. ## 4. AgentMail Raises $6M for AI Email Infrastructure ### Deal Overview - **Stage:** Seed - **Sector:** AI Infrastructure - **Geography:** United States - **Round Size:** $6 million ### Investor Profile General Catalyst and Y Combinator co-led the round with Phosphor Capital. Notable angels include Paul Graham, Dharmesh Shah (CTO of HubSpot), Paul Copplestone (CEO of Supabase), and Karim Atiyeh (CTO of Ramp). Paul Graham’s personal investment is a strong signal — he rarely invests outside of YC’s standard terms. ### Company and Leadership [AgentMail](https://agentmail.to) was founded in 2025 by Haakam Aujla (ex-Optiver quant researcher), Michael Kim (ex-NVIDIA autonomous vehicles), and Adi Singh (ex-Accel, StepStone Group, Flex Capital). The six-person team is based in San Francisco and part of Y Combinator’s S25 batch. ### Problem and Opportunity AI agents need to send emails, receive confirmations, sign up for services, and manage threaded conversations — but traditional email APIs like SendGrid and Postmark were never designed for autonomous, non-human users. Agents need their own inboxes, the ability to self-provision addresses, and infrastructure that supports machine-speed communication patterns. ### Product and Technology AgentMail provides API-based email infrastructure purpose-built for AI agents. Agents can create their own email inboxes, send and receive messages, with full support for two-way conversations, parsing, threading, labeling, searching, and replying. An onboarding API allows agents to self-provision email addresses. The platform now serves tens of thousands of human users, hundreds of thousands of agent users, and over 500 B2B customers. ### Use of Proceeds and Vision The funding will expand the engineering team and platform, accelerating developer and agent adoption. AgentMail is positioning itself as the Twilio of the AI agent era — essential communication infrastructure that every agent stack will need. ### Market Context The AI agent infrastructure market is nascent but growing explosively. AgentMail’s user count tripled when OpenClaw launched in late January 2026 and quadrupled again in February. With 500+ B2B customers and hundreds of thousands of agent users, network effects are already creating platform lock-in. ## Lessons for Founders - **The AI agent economy needs its own infrastructure stack.** AgentMail (email), Manifold (security), and RunSybil (testing) all exist because AI agents create requirements that human-designed tools cannot meet. Founders should look for other human-era infrastructure categories ripe for agent-native rebuilds. - **Deep domain credibility unlocks outsized seed rounds.** RunSybil’s founders came from OpenAI and Meta security; Manifold’s team built the most popular open-source LLM firewall; Claros’s team fabricated working chip prototypes before raising. Investors are paying premiums for teams that have already proven they can build in their domain. - **Hardware moats are back.** Claros’s $30 million seed for chip-level power infrastructure shows that investors are willing to fund capital-intensive hardware when it solves a problem software cannot. The AI infrastructure bottleneck is increasingly physical, not digital. - **Angel investors as customer signals matter.** RunSybil’s angel list reads like a who’s who of security buyers. AgentMail’s angels are CTOs of companies that will integrate the product. Strategic angels who double as future customers or channel partners are more valuable than brand-name checks alone.