Est.

How to Evaluate a Robotics Insurance Broker Before Signing Your First Enterprise MSA

A good robotics broker can match your MSA's insurance exhibit to actual policy language.

· 8 min read
Cover illustration for “How to Evaluate a Robotics Insurance Broker Before Signing Your First Enterprise MSA”
MSA Insurance Requirements · November 16, 2026 · 8 min read · 1,832 words

A robotics company should evaluate its insurance broker against the insurance exhibit in its first enterprise MSA before signing. A skilled broker can match each requirement to policy language, find exclusions that affect autonomous operations, and explain what coverage applies when software causes physical harm. Limits alone do not establish that the company has the coverage its contract requires.

Why the enterprise MSA is the first real test of a robotics broker's competence

A founder receives a draft enterprise MSA, opens the insurance exhibit, and cannot tell whether the current broker can satisfy it. The document asks for more than proof of insurance. It sets conditions for deploying machines at a customer’s site, where autonomous decisions can expose several parties to the same loss. The founder needs to know whether the insurance program supports the contract the company is about to sign.

That makes the exhibit different from a standard commercial policy renewal. The enterprise customer’s legal and risk team has written requirements to cover the liabilities it fears, including software-induced physical loss, third-party bodily injury from machine behavior, and disputes over who caused an incident. A broker must connect them to coverage provisions and secure compliant limits. Failure can leave the robotics company unable to execute the contract or exposed to a breach when a claim reveals that the promised coverage is absent.

What insurance clauses should a robotics startup watch out for in its first enterprise MSA

Enterprise and government MSAs for autonomous systems often call for several coverage lines that a generic commercial program may not have. A technology services MSA usually calls for technology errors and omissions coverage and cyber liability, then adds general liability for bodily injury and property damage from general business operations. With robotics, what a machine does can also cause physical harm. Those exposures can mean the required limits are far above a standard commercial general liability policy’s default limits. Umbrella or excess coverage is added to the program when an enterprise site owner calls for higher aggregate liability limits.

An exhibit may require inland marine or equipment coverage that follows the robot to the customer’s premises. Standard property policies stop at the insured’s own premises, but deployed robots spend much of their working lives elsewhere. The distinction is easy to miss in a placement built around the company’s office that overlooks its operations. A program can satisfy a liability requirement yet leave the deployed equipment without the required protection.

Other clauses govern the terms on which coverage must extend to the customer. These include additional insured status, waiver of subrogation, and notice periods, each of which needs attention beyond the headline limit. Some MSAs also expressly name autonomous or AI-operated systems and require coverage to extend affirmatively to those operations. Silence is not the same as affirmative coverage. The broker must distinguish requirements the current program satisfies from those requiring a different placement or negotiation.

The AI exclusion wave that makes a generic policy dangerous for autonomous operations

The Insurance Services Office introduced optional generative-AI exclusion endorsements for general liability in 2026, including forms CG 40 47, CG 40 48, and CG 35 08. Parallel exclusion language is spreading into directors and officers and errors and omissions forms. These are not merely changes in terminology. When an AI exclusion applies, it can remove coverage for the autonomous operation that causes a loss, even if the policy meets the MSA’s stated limits.

That problem grows when each policy cuts back its AI protections on its own. A single robotics incident may involve software failure, physical damage, and professional liability, while cyber, tech E&O, and general liability policies address different parts of the exposure. Their exclusions may leave an unfilled gap. A tech E&O or cyber policy can appear to satisfy the exhibit’s line-item checklist while excluding the company’s central autonomous exposure.

This fragmentation also weakens reliance on “silent” AI coverage, under which existing policies implicitly covered robotics risks without expressly naming AI. Carriers’ introduction of AI-specific exclusions has accelerated the move away from that position. Past placement knowledge is no longer enough. A broker who checks for changes to exclusions each week can tell affirmative protection from wording that stays silent. For a robotics company, that distinction belongs in the compliance review, before a claim occurs.

Why physical AI creates liability disputes that standard underwriting submissions cannot describe

An autonomous incident can implicate the hardware maker, AI model supplier, and operator at once. Each may bear part of the responsibility, yet a standard commercial underwriting form cannot accurately describe that chain of attribution. Calling the applicant a “technology company” obscures the risk. A useful submission instead describes the machine, its operating environment, its decision logic, and the contractual chain of responsibility. Without that detail, the insurer is being asked to price an operation it does not understand.

The difficulty extends beyond identifying the parties. The same robot can present different risks at two warehouse sites because floor layouts, human interaction patterns, and software versions diverge after the policy is written. The machine, its environment, sensor capabilities, safety controls, and software versioning together explain the exposure that the insurer is accepting. A broker who translates those variables into a coherent submission gives the underwriter a basis for better terms than a generic description provides.

Operational evidence is part of that description. System observability and data logging have become prerequisites for insurability with specialist carriers because high-fidelity logs provide a forensic foundation for pricing risk. Companies without that evidence face exclusions or prohibitive premiums. The cyber-physical boundary adds another complication, since a compromised robot may injure people or damage property. When cyber and physical exposures are treated as separate, the cyber policy may exclude physical consequences while the general liability policy excludes software causation.

The specific questions to ask a broker before signing — and what the answers reveal

Before signing, a founder should test those skills with questions that ask the broker to explain policy language and claims coordination as well as confirm access to insurance. Start by asking whether the broker has read AI exclusions, specifically ISO CG 40 47 and CG 40 48, and which carriers offer affirmative coverage for autonomous operations. A competent broker can describe which carriers in their market use the exclusions and show how affirmative protection differs from silence. A broad assurance that a policy “covers AI” does not answer that question.

Next, ask the broker to walk through a bodily injury claim that the customer attributes to the robot’s software. The broker should explain which policy responds first and what happens if the cyber and general liability carriers dispute the cause. This tests the program at its seams. The answer should explain which insuring agreement reaches the loss, how general liability, tech E&O, and cyber coordinate, and whether an exclusion leaves an exposure that the other policies do not cover. Reciting the coverage lines is not an explanation of their interplay.

The MSA itself provides the next test. Ask how long the broker needs to assess the current program against the exhibit. Ask what the review will examine. A sound response explains how the broker checks each line for limits, additional insured status, waiver of subrogation, and notice periods. It should also prompt questions about the customer’s industry and site conditions. A promise to “take a look” establishes neither the scope of the review nor the broker’s ability to recognize an autonomous-operations requirement.

Ask how coverage follows robots after they leave the company’s building, including during pilots and deployments at customer locations. A broker who suggests standard property coverage without bringing up inland marine has failed to account for where the assets work. That omission is distinct from a liability gap and cannot be resolved by increasing liability limits.

Then ask how the broker would characterize the company in an underwriting submission and what recent autonomous-operations placements demonstrate access to the specialist market. The answer should explain how the submission presents the machine, site, software, safety controls, and allocation of responsibility. A broker unable to describe that process or identify the specialist market segment they access offers little basis for expecting a placement built for machines in the field.

Finally, ask how the broker is paid and ask for written commission disclosure before binding. Transparency is a reasonable baseline for the engagement. Broker commissions are already part of the premium whether a generalist or specialist places the policy, so the commission structure alone is no reason to accept generic coverage. This question tests commercial transparency, which must be considered alongside technical competence because neither can substitute for the other.

What a competent broker actually does differently when placing robotics coverage

A specialist broker’s placement work begins with an account of how the machines operate. The submission describes the operating environment, decision architecture, human interaction model, software versioning cadence, and safety controls in terms an underwriter can price. Specificity changes the basis of the assessment. Physical AI presents limited claims history and high environmental variability, so a generic submission encourages worst-case assumptions or generic pricing adjustments.

A skilled broker reads the policy before binding to check for AI exclusions, autonomous-operations carve-outs, and gaps in how the coverage lines work together. The carrier’s summary is not sufficient evidence. Endorsements may limit what that summary seems to promise, especially when autonomous operation links software causation to physical harm.

To check MSA compliance, the broker matches each requirement in the exhibit to the policy provision that meets it. That check covers additional insured endorsements, waiver of subrogation, notice-of-cancellation periods, and exactly how the limits are structured. Unsupported requirements are flagged before the contract is signed. The placement also coordinates general liability, tech E&O, cyber, inland marine, and umbrella coverage around the company’s enterprise deployment. These policies have different jobs, but their interaction must account for an incident that crosses more than one line.

How to use the MSA insurance exhibit as a live vetting document before you sign anything

The practical next step is to send the exhibit to the broker before signing and request a written gap analysis against the current program. Treat the response as an evaluation. The requested document should provide line-by-line commentary showing what the exhibit requires, what the current policy provides, where a gap exists, and what placement would close it. This format makes the broker’s reasoning visible. It also separates a requirement already supported by policy language from one that still depends on further work.

A reassurance without that analysis leaves the founder unable to tell whether the broker has spotted affirmative autonomous-operations language, problems with how the coverage lines work together, or equipment exposure away from the company’s premises. The written response should make those distinctions explicit. If it instead reveals unfamiliarity with AI exclusions, autonomous operations, or inland marine for field-deployed hardware, the signal is to find a specialist before signing. The decision belongs at the contract stage, while a coverage gap can still be resolved before a loss turns it into a dispute.

Sources

  1. Robotics, Third-Party Integration, and the New Risk Frontier

More in MSA Insurance Requirements