US Seed Weekly Funding Roundup (Apr 27-May 4, 2026): $72.1M Raised Across 4 Deals

Table of Contents

US seed-stage funding totaled $72.1M across 4 deals between April 27 and May 4, 2026. Three of the four winners are agentic AI companies, and the fourth is a defense-focused AI integrator. The smallest check on the list was $6M, the largest $40M, and every round closed with a tier-one investor on the cap table.

The week looked thin compared to later-stage activity in the same geography, where Series A rounds pulled in $378M across 7 deals and Series B+ companies raised $1.12B across 7 more. That gap reflects how capital is concentrating right now: investors are willing to bet small at the frontier and large at proven scale, with less appetite for the middle. Each of the four seed rounds below targets a workflow that legacy software hasn't cracked.

1. NeoCognition Raises $40M For Self-Learning Enterprise Agents

Deal Overview

• Stage: Seed
• Sector: AI / agentic infrastructure
• Geography: Palo Alto, United States
• Round size: $40M
• Lead investors: Cambium Capital, Walden Catalyst Ventures, Vista Equity Partners, with angel checks from Lip-Bu Tan and Ion Stoica

Investor Profile

Walden Catalyst and Vista Equity bring enterprise distribution muscle, while Lip-Bu Tan (Cadence, Intel) and Ion Stoica (Databricks, Anyscale) signal infrastructure and academic credibility. That's an unusually deep bench for a seed round and tells you something about how investors view the founders' technical lead.

Company and Leadership

The team is roughly 15 people, mostly PhDs, led by AI researchers Yu Su, Xiang Deng, and Yu Gu out of the Ohio State NLP group. Their prior academic work includes Mind2Web and other web-agent benchmarks that defined how the field measures agent capability.

Problem and Opportunity

Today's enterprise AI agents are largely prompt-engineered and brittle. They don't learn from their own execution inside a customer's environment, so every new workflow requires another round of human tuning. Buyers want agents that get sharper the longer they run, not ones that need to be rebuilt every quarter.

Product and Technology

NeoCognition is building "expert agents" that acquire a world model of work by observing the systems and constraints around them. The thesis: trajectories collected during real agent execution compound into a specialization moat that horizontal foundation models can't replicate.

Use of Proceeds and Vision

The $40M funds research and engineering hires, training infrastructure for self-learning agents, and design partnerships with SaaS incumbents and Fortune 500 buyers. The pitch sits between horizontal foundation labs like OpenAI and vertical agent SaaS players like Sierra and Decagon.

Market Context

Enterprise agent funding has been one of 2026's loudest categories. NeoCognition's bet on continual learning differentiates it from the prompt-orchestration crowd, and the angel list buys it credibility with infrastructure and enterprise buyers from day one.

Case Studies

Startups like yours already closed their rounds with us.

Founders across every stage and industry. Here's what it took.

• Raised $7.6M for Swiipr Technologies
• Raised $0.5M for Ap Tack
• Raised €0.5M for Ivent Pro

Read their stories →

2. Rilian Raises $17.5M For Agentic Defense Cybersecurity

Deal Overview

• Stage: Seed (combined with seed extension)
• Sector: Cybersecurity / defense tech
• Geography: McLean, Virginia, United States
• Round size: $17.5M
• Lead investors: 8VC, First In, Tamarack Global, with 8090 Industries, Liquid 2 Ventures, Perot Jain, and Protego Ventures participating

Investor Profile

8VC and First In are both active defense-tech backers, and Tamarack Global has a track record in dual-use deep tech. The syndicate is built for federal procurement, not consumer software.

Company and Leadership

Rilian was founded in 2024 and is based in McLean, Virginia, a deliberate choice given its government and critical-infrastructure focus. The company already counts the UAE Cybersecurity Council as a customer through a July 2025 contract.

Problem and Opportunity

Legacy SOAR playbooks break down in air-gapped, on-premises, and compliance-heavy environments where commercial cloud-only tools can't operate. Sovereign defense and critical-infrastructure buyers need automation that works inside their constraints, not around them.

Product and Technology

Rilian's flagship platform Caspian unifies cyber and defense tooling across cloud, on-premises, and air-gapped environments. It deploys AI agents trained to assess risks and respond autonomously, replacing rigid playbooks with model-driven workflows.

Use of Proceeds and Vision

Funds will accelerate go-to-market with allied governments and critical-infrastructure customers, expand AI research talent, and deepen R&D in agentic defense workflows. The UAE National SOC deployment is the reference case Rilian plans to extend into other sovereign markets.

Market Context

Defense tech and AI-native cyber are seeing record VC inflows, with True Anomaly's $600M space-security round topping the same week's funding leaderboard. Demand is driven by escalating nation-state cyber activity and government willingness to procure from venture-backed AI vendors over legacy primes.

3. Iridius Raises $8.6M For Compliance-By-Design AI

Deal Overview

• Stage: Seed
• Sector: AI / regulated workflow infrastructure
• Geography: United States
• Round size: $8.6M
• Lead investor: Chalfen Ventures, with Osage Venture Partners, Accenture Ventures, and Rock Yard Ventures participating

Investor Profile

Accenture Ventures matters here because Accenture is one of the largest systems integrators selling into life sciences and pharma. Osage and Chalfen bring enterprise SaaS pattern recognition.

Company and Leadership

Iridius is led by CEO Mike Kropp and was founded by Microsoft and AWS veterans. Its advisory board reads like a pharma executive directory: Pfizer, Merck, Bayer, Novartis, Johnson & Johnson, Medtronic, Alexion, and Microsoft.

Problem and Opportunity

Regulated industries account for trillions in global GDP, but AI adoption has been throttled by manual validation, fragmented compliance, and review processes that don't scale. Buyers can't deploy agents into GxP workflows if they can't generate audit evidence by construction.

Product and Technology

Iridius embeds compliance directly into AI systems as executable logic, enforcing rules continuously and generating evidence across the workflow lifecycle. Regulatory requirements like GxP become first-class primitives that agents must satisfy, rather than review steps bolted on after the fact.

Use of Proceeds and Vision

Seed capital funds product build-out, regulatory and engineering hires, and design-partner deployments. The roadmap extends the policy-as-code library beyond GxP to adjacent regimes in finance, healthcare, and government.

Market Context

Competitors include Veeva on the life-sciences side and emerging GRC-AI startups like Vanta, Drata, and Norm AI. Iridius's executable-compliance approach and pharma advisory base give it a different wedge than horizontal GRC tools.

4. Balerion AI Raises $6M For Mortgage Origination Agents

Deal Overview

• Stage: Seed
• Sector: AI / fintech (mortgage)
• Geography: United States
• Round size: $6M
• Lead investor: Kleiner Perkins, with Formation and BoxGroup participating

Investor Profile

Kleiner leading a $6M seed is a signal in itself; the firm typically anchors larger checks. Formation and BoxGroup add early-stage pattern recognition in vertical AI.

Company and Leadership

Balerion AI emerged from stealth in April 2026 and is already deployed with FM Home Loans, a residential lender managing more than $2B in loan volume. Production traction at launch is rare for a seed company.

Problem and Opportunity

US mortgage origination costs roughly $12,000 per loan and takes weeks to close. Lender margins have been squeezed since 2023, and most automation tools attack one task at a time, leaving the rest of the file to humans.

Product and Technology

Balerion Loan Intelligence is an agentic platform covering the full loan manufacturing lifecycle, from intake through underwriting and closing. It calculates income, flags derogatory items in bank and credit data, and enforces compliance with Fannie Mae, Freddie Mac, FHA, and non-QM overlays. Signals are shared across the entire file rather than siloed by task.

Use of Proceeds and Vision

Funds will deepen the platform's AI capabilities, expand enterprise customers among lenders, and grow engineering and go-to-market teams. The longer-term goal is to compress origination cost and timeline by an order of magnitude.

Market Context

Legacy LOS incumbents like Encompass and Blend are the targets, and the agentic-mortgage category is filling fast with rivals including Floify AI, Tavant, and several Y Combinator entrants. Production deployment with a $2B+ lender at seed is a meaningful early lead.

Lessons For Founders

• Tier-one angel investors at seed (Lip-Bu Tan, Ion Stoica, Andrej Karpathy in adjacent rounds) are increasingly the differentiator between a $6M and a $40M check. Build the angel list before you raise.
• Three of four winners this week ship agentic AI. The pattern: own the full workflow, not a single task. Single-point AI tools are losing to platforms that share signals across the file.
• Production traction at launch beats stealth narrative. Balerion closed Kleiner with a live $2B+ lender deployment; Rilian closed 8VC with a sovereign customer already paying.
• Regulated and defense buyers reward depth over speed. Iridius assembled a pharma advisory board before raising; Rilian set up in McLean. Geography and governance signal seriousness to procurement.
• If you're building horizontal AI, expect to compete with foundation labs. The seed winners this week all picked verticals deep enough that horizontal models can't follow without rebuilding the data flywheel.