Saronic Technologies: Engineering Autonomous Surface Vessels for the U.S. Navy
How a defense maritime startup navigates undefined standards, contested environments, and the long road from prototype to program of record
The Moment Saronic Is Building Into
Austin-based Saronic Technologies has attracted significant attention — and significant funding — by pursuing one of the more technically demanding problems in defense: building autonomous surface vessels (ASVs) capable of operating in contested maritime environments, without continuous human oversight, at a cost point that makes them tactically expendable.
The company’s publicly disclosed vessels — the Defend-class and Corsair-class platforms — are designed to operate as part of distributed maritime forces, conducting ISR, logistics, and, potentially, strike-relevant missions. The broader concept of operations they fit into, the Navy’s Distributed Maritime Operations (DMO) doctrine, envisions swarms of lower-cost autonomous and semi-autonomous assets extending the sensor and effects range of manned surface combatants.
That’s the strategic vision. The engineering reality is considerably messier. Saronic is building complex systems at the intersection of three sets of demands that don’t yet have a coherent shared framework: maritime safety regulations designed for crewed vessels, military acquisition standards designed for conventional platforms, and emerging autonomy assurance frameworks that remain incomplete for any vehicle class. Understanding how those demands interact — and where they conflict — is essential to understanding the systems engineering challenge Saronic and companies like it face.
The Standards Problem: Four Regimes, No Single Authority
Every naval architect working on a crewed commercial vessel knows the regulatory landscape: COLREGS (the International Regulations for Preventing Collisions at Sea), classification society rules, and flag state requirements. These are mature, well-understood, and extensively documented.
None of them were written for unmanned systems.
COLREGS Rule 2 assigns responsibility to the “master or owner” for departures from the rules when the ordinary practice of seamen requires it. Rule 5 requires a proper lookout “by sight and hearing.” These provisions are not technically incompatible with autonomous operation, but they create genuine ambiguity about where legal and operational responsibility sits when a vessel’s decision-making is algorithmic. The IMO has been working on a Maritime Autonomous Surface Ships (MASS) regulatory framework since 2017, with a goal-based instrument expected to enter into force no earlier than 2028. That timeline provides cold comfort to a company trying to field systems today.
Military standards introduce a parallel set of requirements. Naval Sea Systems Command (NAVSEA) acquisition frameworks, the JCIDS (Joint Capabilities Integration and Development System) process, and relevant MIL-SPECs governing software-intensive systems (MIL-STD-498 successors, DO-178C analogs for defense-critical software) all apply in various degrees. These standards are designed for programs with multi-year development cycles, dedicated program offices, and established contractor relationships. A startup operating on venture-backed timelines is not the intended user.
The third regime is the newest and, in some ways, the most demanding. DoD Directive 3000.09 on autonomous weapons systems — and the AI assurance requirements flowing from it — imposes requirements for human oversight, testing, and operational constraints on lethal autonomous systems. Whether Saronic’s platforms fall under this directive depends heavily on their intended operational use, a question that will be answered differently across different program contexts. The DoD AI Ethical Principles and the Responsible AI (RAI) framework add additional layers, requiring explainability, traceability, and bias testing for AI-enabled decision-making — concepts that are conceptually clear but operationally underdefined for maritime autonomy specifically.
The fourth regime is the Navy’s own emerging policy on unmanned systems. The 2023 Unmanned Campaign Framework and associated implementation guidance provide directional intent — more unmanned, distributed, affordable — but leave substantial gaps in the acquisition mechanics for novel platforms.
The practical consequence: Saronic’s systems engineering team is not working against a single authoritative requirements baseline. They are simultaneously negotiating interpretations across multiple partially overlapping frameworks, some of which are actively evolving as they build.
Communication-Denied Operations: The Autonomy Requirements Problem
The tactical value proposition of Saronic’s platforms depends heavily on their ability to operate in communication-degraded or denied environments. This is not a peripheral capability — it is core to their relevance in a conflict scenario against a near-peer adversary with demonstrated electronic warfare and anti-satellite capabilities.
Operating without reliable communication links means the vessel’s autonomy stack must make real-time decisions about navigation, threat assessment, rules of engagement boundaries, and mission abort criteria without human-in-the-loop guidance. This creates a requirements problem that has no clean precedent.
For conventional weapons systems, the requirements for autonomous operation are narrow and well-defined: a missile’s terminal guidance is autonomous, but its target selection happens before launch. For Saronic’s platforms, the scope of autonomous decision-making is much broader and context-dependent. A vessel executing a surveillance mission in a congested maritime environment may need to make dozens of independent judgments about vessel identification, rules of engagement compliance, sensor management, and self-preservation over the course of a single sortie.
Writing requirements for this behavior is genuinely hard. The challenge is not primarily technical — the autonomy algorithms exist, and they continue to improve. The challenge is specifying what the system shall do in a way that is verifiable, unambiguous, and defensible in a DoD acquisition context. How do you write a shall-statement for graceful degradation of mission objectives under progressive communications loss? How do you define verification criteria for a behavior that depends on environmental context that cannot be fully enumerated in a test plan?
These are active problems in the defense autonomy community, and they do not yet have accepted answers. What they require, at minimum, is a requirements architecture that can represent conditional and adaptive behaviors — not a flat list of text requirements — and a traceability framework that connects those behaviors to verification evidence, test scenarios, and safety arguments.
From Prototype to Program of Record: The Systems Engineering Inflection Point
Saronic has demonstrated functional hardware. Corsair-class vessels have been publicly shown operating in the Gulf of Mexico, and the company has participated in Navy exercises. Getting a vessel to operate autonomously in controlled conditions is a significant engineering achievement. It is not the same thing as being ready for a Navy program of record.
The transition from prototype to program of record is, in many respects, a systems engineering transition rather than a technology transition. The questions that dominate are not primarily about whether the system can do the mission — they are about whether the program can prove it, maintain it, evolve it, and support it at scale.
A Navy program of record will require, at a minimum: a complete and traceable requirements baseline from mission-level needs through system, subsystem, and component requirements; a verification and validation plan with evidence for each requirement; a configuration management system capable of managing changes across hardware, software, and firmware with full audit trails; a test and evaluation record that satisfies both developmental testing (DT) and operational testing (OT) criteria; and a technical data package sufficient to support sustainment and competitive procurement of components.
None of these are insurmountable for a well-organized company. All of them require early investment in processes and tooling that may not feel urgent when the immediate priority is getting the vessel to do something impressive in front of a customer. The companies that reach program of record will be those that built systems engineering discipline into their development process early enough that the documentation is a natural output of their work, not a retroactive reconstruction of what they built.
This is where the choice of requirements management and systems engineering tooling matters more than it might appear. Legacy tools — document-centric approaches that store requirements as formatted text in isolated repositories — create significant overhead when programs scale. They make traceability manual, change impact analysis error-prone, and verification tracking difficult to automate. Modern platforms built on graph-based models, where requirements, design elements, test cases, and verification evidence are nodes in a connected structure, make the artifacts that DoD acquisition processes demand a direct product of the engineering workflow rather than a separate documentation effort.
Flow Engineering, for example, is built specifically for this kind of connected systems engineering work — linking requirements to design rationale, tracking verification status in real time, and supporting the kind of change impact analysis that becomes critical when a requirement changes and you need to immediately understand everything downstream that is affected. For a company like Saronic, where requirements will evolve across multiple customer engagements and hardware generations, that kind of infrastructure has direct bearing on how efficiently they can respond to DoD source selection data calls.
The Competitive and Strategic Context
Saronic is not alone in this space. Textron’s CUSV (Common Unmanned Surface Vehicle), L3Harris, Elbit, and several other primes and non-traditionals are pursuing overlapping capability areas. The Navy’s stated intent to significantly expand its unmanned surface vessel inventory — including a Ghost Fleet Overlord program that has already demonstrated multi-vessel autonomous operations — creates a real market, but also a competitive one.
The primes bring existing program relationships, established compliance infrastructure, and NAVSEA familiarity. Non-traditional entrants like Saronic bring faster iteration cycles, modern software architectures, and engineering cultures less constrained by how things have been done before. Neither set of advantages is permanent, and neither is sufficient on its own.
The companies that will win programs of record will combine the technical capability to field systems that meet mission requirements with the organizational maturity to satisfy the acquisition rigor that DoD programs demand. That combination — fast iteration plus disciplined systems engineering — is genuinely difficult to achieve simultaneously. It requires deliberate choices about process, tooling, and engineering culture from early in the program lifecycle.
Honest Assessment
Saronic is working on a real capability gap at a moment when the Navy has both the doctrinal intent and, at least directionally, the budget intent to close it. The technical challenges — autonomous navigation, communications-denied operations, multi-vessel coordination — are hard but tractable. The standards and acquisition challenges may be harder, because they require not just engineering solutions but the institutional infrastructure to demonstrate those solutions in the formats and with the evidence that defense acquisition processes require.
The open question is not whether autonomous surface vessels will become a significant part of U.S. naval force structure. The open question is which companies will have built the systems engineering foundations necessary to survive the transition from demonstrator to program. That transition rewards companies that treated requirements management and traceability as first-order engineering problems rather than compliance afterthoughts.
The maritime autonomy race is as much about engineering process as it is about engineering product.