AgentStatus × CFC
Independent, distributed assurance for production AI agents, alongside CFC's Lane Assist.
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 Lane Assist and the broader AI capabilities CFC is building across cyber and specialty insurance. We do not replace them.
What we understand about CFC
A specialty insurance pioneer running the world's first agentic underwriting pilot.
CFC is the specialist insurance provider, pioneer in emerging risk and market leader in cyber, headquartered in London with offices in New York, Austin, Brussels, and Brisbane. Over 700 staff. More than 130,000 businesses insured across 90+ countries. Backed by Lloyd's, regulated by the Financial Conduct Authority, and one of the largest independent managing general agents in the world.
In April 2026, CFC launched Lane Assist, described publicly as a world-first pilot of agentic underwriting in specialty insurance, taking a submission from email through to quote recommendation in seconds. The pilot is live in CFC's cyber underwriting team, processing real new business submissions, with every agent recommendation reviewed and approved by an underwriter before issuance.
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.
Underwriter approval vs continuous evidence
Pilot validation vs production drift
Global execution footprint
Partner-friendly integration posture
The split
How the work divides
How the work divides
Their platform
- • Lane Assist agentic underwriting
- • Underwriter-approved recommendations
- • CFC underwriting rules & practices
- • Cyber & emerging risk specialty
- • FCA-regulated, Lloyd's-backed
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."
We have also caught node operators trying to game the network with datacenter VMs instead of real consumer egress. Detection of adversarial behaviour is built into the product. 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 Lane Assist, the underwriter approval workflow, or the broader AI capability CFC is building across cyber and specialty insurance. We are an independent layer that can coexist with them, and where useful, help CFC correlate outside-in validate outcomes with inside-out underwriter approval data, so the AI team has continuous evidence the deployed agent is still behaving the way CFC's rules require, as the pilot scales.
What we'd like from this conversation
These three asks would move a pilot forward.
Validate the fit
As Lane Assist scales beyond the cyber pilot, where does CFC want independent assurance, and where does CFC prefer everything native to the underwriter-in-the-loop workflow?
A practical next step
A sandbox Lane Assist surface we can validate with gold prompts representative of a cyber or specialty submission, so CFC's agentic underwriting and AgentStatus drift detection tell one story together, particularly useful when the pilot expands beyond low-complexity cyber risks.
Partner path
If there is a partner path, we'd like to understand supported integration patterns as CFC's AI capability scales, particularly given the divergence between the EU AI Act's codified regime (operating in Brussels) and the UK's principles-based approach (operating in London).
CFC × AgentStatus.
CFC helps brokers and customers build and operate the world's first agentic underwriting pilot in specialty insurance. AgentStatus helps CFC prove, continuously, that the deployed agent behaves the way underwriters, brokers, and the FCA 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 CFC references above reflect public product pages, the Lane Assist announcement (April 2026), and CFC's regulatory disclosures as of the date of this note.