AgentStatus × Cytora - a quick map of how we fit
Independent, distributed assurance for production AI agents, alongside Cytora's agentic risk processing platform.
Two jobs: reachability from residential networks (past CDN/WAF), then reliability once reached — gold/contract and consistency checks, with a scoped ceiling on judges. We sit alongside your platform. We do not replace them.
What we understand about Cytora
The pioneer of agentic AI applied to commercial insurance workflows.
Cytora is the digital risk processing platform built for commercial insurance, with the Risk Flow Engine at the core: a low-code platform to define and execute multi-step, human-in-the-loop risk processing workflows across digitize, evaluate, and decide stages. The Underwriter Console surfaces decision support information, while the Human-in-the-loop Console intelligently routes low-confidence outputs to human operators.
In March 2026, Cytora launched Autopilot, a major agentic AI capability that enables underwriting and claims workflows to run themselves end-to-end. Autopilot aggregates and interprets internal, external, and submission data across emails, documents, and calls, executes autonomously as the picture of a risk evolves, and provides explainable agentic reasoning where every workflow step is fully auditable with transparent reasoning records.
Cytora customers include Chubb (digitizing global Claims document flows), Markel (underwriting digital risk flows), and Arch Insurance (recently expanded to London Market operations). The platform is part of the Applied Systems portfolio, with strategic data partnerships across LexisNexis Risk Solutions, The Warren Group, Altitude Intelligence, and Google Cloud. London-headquartered. Named winner in the 2026 AI Excellence Awards, Insurance product category.
What AgentStatus is
We measure whether users can reach the agent, then whether it still passes its checks.
Reachability. Controlled validate traffic from 2,500+ residential devices across 70 countries measures whether users can open the agent the way they do — past CDN, WAF, and bot walls that treat datacenter synthetics differently. Multi-geo is observer vantage for access and last-mile latency. It does not change agent tool egress or localize answers by probe IP.
Reliability. Once reachable, we run gold/contract checks when truth exists, plus rephrase, drift, and policy consistency probes when it doesn't. Dual LLM-as-judge scores open answers with a known ceiling — stably wrong but consistent still needs a domain expert.
Outside-in validation is two separate jobs
Reachability
Outcome
Residential path
Your monitor hits the VIP lane. Users hit the WAF. Datacenter checks get blocked, throttled, or allowlisted. Residential observers take the inbound path customers take — so “up” means reachable from home networks, not from AWS.
Reliability
Outcome
Answer quality
Reachable and self-contradicting is still broken. Rephrase flips, drift, and policy breaks need no ground truth. Gold and dual judges cover the rest when truth exists. Uptime grades none of that.
Where we fit
We sit beside the platform. We do not replace it.
Auditable reasoning vs continuous evidence
Human-in-the-loop vs production drift
Global execution footprint
Partner-friendly integration posture
The split
How the work divides
How the work divides
Their platform
- • Risk Flow Engine
- • Cytora Autopilot agentic AI
- • Underwriter & HITL Consoles
- • Explainable reasoning records
- • Applied Systems / Cytora ecosystem
Outcome
System of record
Dashboards, exports, lifecycle tools, and orchestration remain theirs. We do not replace that surface.
AgentStatus
- • Continuous validate traffic
- • Gold libraries & drift detection
- • Multi-turn / multi-agent journeys
- • Real-network execution evidence
- • 2,500+ nodes across 70 countries
Outcome
Outside-in layer
Residential inbound path past CDN/WAF, then gold, consistency, and scoped judges once the agent is reachable.
Proof of scale
Auditable scale metrics
In about two months, we have executed on the order of 18 million validate runs across the network. We also maintain on the order of 6,000 agent records in our system, meaning rows and configurations we track, including evaluation and pipeline agents, not "6,000 paying customers."
If helpful, we can share stricter production-only definitions under NDA.
What we are not claiming
We are an independent layer that runs alongside your stack.
We are not a replacement for the Risk Flow Engine, Cytora Autopilot, the Underwriter Console, or the Human-in-the-loop Console. We are an independent layer that can coexist with them, and where useful, help carriers correlate outside-in validate outcomes with inside-out reasoning records, so underwriting and claims leaders have continuous evidence the deployed agents are still behaving the way the platform intended, as Autopilot scales beyond the launch cohort.
What we'd like from this conversation
These three asks would move a pilot forward.
Validate the fit
As Autopilot scales across Chubb, Markel, Arch, and the broader carrier base, where do customers want independent assurance, and where does Cytora prefer everything native to the Risk Flow Engine?
A practical next step
A sandbox surface we can validate with gold prompts representative of an underwriting or claims workflow (submission ingestion, enrichment via partner data, risk decisioning, claims intake), so Autopilot's reasoning records and AgentStatus drift detection tell one story together.
Partner path
If there is a partner path, we'd like to understand supported integration patterns for carriers, MGAs, and reinsurers operating across the Lloyd's Market, EMEA, and North America, particularly through Cytora's strategic data alliances and the broader Applied Systems portfolio.
Cytora × AgentStatus.
Cytora helps carriers build and operate the world's first agentic underwriting and claims workflows. AgentStatus helps those same carriers prove, continuously, that the deployed agents behave the way regulators, auditors, and policyholders require, globally, with evidence that holds up under scrutiny.
Contact·dulra@carmel.so·roman@carmel.so
Metrics are stated with explicit definitions: validate runs are scheduled executions over ~two months; agent records are database rows, not revenue customers. Public Cytora references above reflect public product pages, the Cytora Autopilot announcement (March 17, 2026), the Chubb / Markel / Arch / LexisNexis / Warren Group / Altitude Intelligence partnership announcements, and Applied Systems portfolio disclosures as of the date of this note.