AgentStatus × Sixfold
Independent, distributed assurance for production AI agents, alongside Sixfold's underwriting AI platform.
Two jobs: reachability from residential networks (Monitoring now; Reliability over time), then outcome verification once reached - Scenario, Compositional, Safety, Stability (Consistency / Drift scores), not whether it reused the same words. We sit alongside your platform. We do not replace them.
What we understand about Sixfold
The first AI purpose-built for insurance underwriters.
Sixfold's platform brings AI agents directly into the underwriter's workflow, agents that triage cases, surface cited risk insights, score every submission 1 to 5 against carrier guidelines, write referral emails, and audit cases for compliance issues before binding. The latest advancement, Institutional Intelligence, encodes a carrier's risk appetite, underwriting guidelines, and historical decisions so the platform answers the way the carrier expects.
Sixfold operates with insurance-grade data governance and isolation, integrates via API into existing workbenches, CRMs, and policy admin systems, and is deployed in production at carriers including Zurich North America, Generali Global Corporate & Commercial, Skyward Specialty, and Mosaic. Backed by a Series B from Brewer Lane Ventures, with strategic investment from Guidewire Software, Bessemer, and Salesforce Ventures.
What AgentStatus is
We measure whether users can reach the agent, then whether it did the job.
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. Monitoring asks if it is healthy right now; Reliability asks if it keeps working over time. Multi-geo is observer vantage for access and last-mile latency. It does not change agent tool egress or localize answers by probe IP.
Outcome verification. Once reachable, we verify outcomes: Scenario (did it finish the job?), Compositional (do the pieces hold together?), Safety (must-not-say / policy / attacks), Stability (same ask, same story?). Consistency and Drift track what changed. Job anchors and side-effects where they exist — not prose matching. Optional sample review is corroboration only; stably wrong still needs a domain expert.
User-side validation is two separate jobs
Reachability
Outcome
Can we talk to it?
Residential observers take the inbound path customers take — past CDN, WAF, and bot walls that treat datacenter synthetics differently. Monitoring asks: is it healthy right now? Reliability asks: does it keep working over time? “Up” means reachable from home networks, not from AWS.
Outcome verification
Outcome
Did it do the right thing?
Reachable and wrong is still broken. Scenario — did it finish the job? Compositional — do the pieces hold together? Safety — must-not-say, policy, attack probes. Stability — same ask, same story? Scores: Consistency and Drift. Not prose matching. Optional sample review is corroboration only.
Where we fit
We sit beside the platform. We do not replace it.
Eval-time accuracy vs production drift
Inside-out scoring vs user-side evidence
Global execution footprint
Partner-friendly integration posture
The split
How the work divides
Your platform
- • Institutional Intelligence
- • 1-5 appetite-fit scoring
- • Cited risk insights
- • Referral & compliance agents
- • Workbench / CRM / PA integration
Outcome
System of record
Dashboards, exports, lifecycle tools, and orchestration remain yours. 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
User-side layer
Reachability (Monitoring / Reliability) from residential networks, then outcome verification once reached — Scenario, Compositional, Safety, Stability; Consistency and Drift scores. Not prose matching.
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 Sixfold's Institutional Intelligence, agent library, or workbench integrations. We are an independent layer that can coexist with them, and where useful, help carriers correlate user-side validate outcomes with inside-out scoring accuracy, so underwriting leaders have continuous evidence that the deployed agent is still behaving the way the carrier's appetite requires.
What we'd like from this conversation
These three asks would move a pilot forward.
Validate the fit
Where do Sixfold's carrier customers want independent assurance for agent output, and where does Sixfold prefer everything native to the platform?
A practical next step
A sandbox agent we can validate with gold prompts representative of an underwriting workflow (submission triage, appetite-fit scoring, referral generation), so Institutional Intelligence 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 operating across multiple geographies and lines, particularly through Sixfold's strategic alliances with Guidewire and across the Lloyd's Market.
Sixfold × AgentStatus.
Sixfold helps carriers build and operate AI agents that bring joy back to underwriting. AgentStatus helps those same carriers prove, continuously, that those agents behave the way appetite, regulators, and brokers 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 Sixfold references above reflect Sixfold's public product pages, customer disclosures, and Series B announcement as of the date of this note.