Article available: Why underwriters, not legislators, will set the terms for physical AI
Live

Insure & secure your
Physical AI

The first coverage built for non-deterministic models controlling robots that move, lift, drive, and decide in the real world.

✓ You're on the list — we'll be in touch.

Email only. We'll reach out within a week.

TALOS
▸ COVERED

Legacy policies cover mechanical failure. We insure the non-deterministic gap — the moments an AI agent makes a probabilistic decision that ends in property damage, kinetic injury, or business disruption.

That’s the purpose-built indemnity your team needs to move from pilots to deployment.

Everything you need to secure your robot's deployment, in one place.

◢ Agent Risk Index
Scanning live
High risk
82/100 HIGH

Every deployed agent carries non-deterministic risk. Talos scores it continuously — and caps your downside when it bites.

What we offer

01

End-to-end peace of mind.

One policy from first prototype to global fleet. No gaps in between.

02

Collaborative risk modeling.

Priced on your actual system, from your safety case, ODD, and incident data, not a generic table.

03

Procurement-ready coverage.

Policies that clear enterprise procurement on day one, so insurance never stalls a deal.

04

Scalable fleet protection.

Ten units or ten thousand: new hardware and geographies, no re-brokering.

Talos-1

Our safety standard

Talos-1 is the first safety standard built for operators of physical AI. We score your deployment across five categories, weight them, and that score sets your terms.

  1. 01

    Operational Design Domain

    Where, when, and under what conditions is the system allowed to operate?

    25%
  2. 02

    System Validation

    What evidence shows it is safe inside that envelope?

    25%
  3. 03

    Runtime Monitoring

    While it operates, how do we know it stays in bounds?

    20%
  4. 04

    Change Management

    When the model, hardware, or geography changes, how is it re-validated?

    15%
  5. 05

    Incident Response

    When something goes wrong, who is accountable and what is logged?

    15%
Composite score → terms
< 2.0Decline
2.0 – 2.9Refer
3.0 – 4.4Standard terms
≥ 4.5Enhanced terms

Talos-1 v0.1 — currently in closed development, open-sourcing soon. A low score in any one category can’t be bought back by the others.

Take the Talos-1 self-assessment → ~6 min · instant, self-reported readiness score · no evidence required

What we cover

Coverage for the risks you’re exposed to right now.

Third-party
Third-party bodily injury

Injury to the public, customers, or bystanders from a system in motion.

ODD · Validation
Third-party
Property damage

Damage to third-party equipment, inventory, vehicles, or structures.

Validation
Third-party
Cyber-physical compromise

A hijacked or attacked system that causes physical harm or loss.

Runtime · Incident
First-party
Worker injury

Harm to the operators and staff working alongside the system.

ODD · Runtime
First-party
Product & payload loss

Loss or damage to the cargo, product, or payload the system handles.

Runtime monitoring
First-party
Business interruption

Operator downtime and lost output when a unit or fleet goes down.

Change management

Example coverage

Active coverage for physical AI

A live policy, sized to your fleet and priced by your Talos-1 score. Limits, premium, and terms move with the score.

48-hour quote 5-day bind ODD-bound Re-rated on change

Building physical AI? Join the waitlist and we'll be in touch within a week.

✓ You're on the list — we'll be in touch.