Decision velocity
Queue prioritization, SLA pressure, and recommendation quality are surfaced before underwriters open a blank thread.
ACTION FIRST
<30s
Avg full analysis
9
Specialist agents
18 credits
Typical full run
100%
Traceable outputs
◑ Operator Outcomes
More speed, more confidence, less operational drag.
Risk transparency
Every major claim in the memo is source-backed, so reviewers can validate quickly instead of auditing from scratch.
CITED OUTPUT
Portfolio discipline
Concentration and treaty pressure are evaluated during each submission, not only during end-of-week reporting.
ALWAYS-ON GUARDRAILS
◆ In days, not months
Every Vortic capability ships into your existing book on day one — no replatform, no pilot, no consultant deck.
◑ Flagship workbench capabilities
The two surfaces that change how a portfolio manager spends their morning.
Concentration and CAT response are the two outcomes that used to require a senior analyst, a spreadsheet, and a Monday-morning catch-up. Vortic surfaces both as live workbench views — every poll, every event, every bind.
◆ Concentration, surfaced
Live ZIP3 / postcode aggregation against your CAT cap.
Hex map of every active bound risk. Capital and PML estimates (OEP 1-in-100, OEP 1-in-250, AEP 1-in-10 net, S&P-A charge) re-rolled on every poll. When a ZIP3 crosses the cap, Vortic drafts a rebalance proposal — declines sorted by GWP:TSI, headroom to re-deploy elsewhere, treaty cession refreshed. Never auto-executed.
Open Concentration
◆ CAT response, in minutes
Re-score the entire book against a live hurricane in 18 minutes.
The moment a CAT event is declared, every active policy is scored against the storm track: distance to the eye, wind damage curve, surge fraction, gross loss, treaty cession, net retained. Worst-loss policies surface first. Treaty cession draft and cedent / insured letters are generated as drafts — never auto-sent.
Open CAT mode
◑ Beyond underwriting
Four more surfaces a carrier runs every day.
Claims adjudication, straight-through fast-track binding, a platform builder for integrations + master data + ontology + process discovery, and dynamic sub-agents that the pipeline spawns reflexively when a parent specialist needs a focused second opinion. Built on the same audit-grade agent fabric as underwriting.
◆ Claims & adjudication
FNOL to adjudication memo. With the audit trail every state DOI wants.
A separate top-level surface for adjusters. FNOL parser extracts claim facts from broker email or voice transcript. Coverage Analyst joins against the policy to decide covered / partial / not_covered with cited clauses. Reserve recommender + revision history. Per-adjuster payment authority gating with refer-to-senior. Statutory letters with UCSPA-compliant language for CA / FL / NY / TX. Supervisor queue and book insights.
See claims workflow
◆ Straight-Through Processing
Auto-bind the 60–80% of submissions that don't need a human.
Every submission that finishes the pipeline with all greens, within concentration cap, within pricing band, and outside hard exclusions is auto-bound — bound_risks row created, audit trail stamped, no human touch. Configurable TSI cap, required-green agents, excluded occupancies and states, pricing rate band, post-bind concentration limit. Every STP decision carries full reasons.
See pricing
◆ Platform builder
Configure the platform without a forward-deploy engineer.
Four admin surfaces: Integrations (REST / webhook / database / SFTP / stream / S3 connectors with auth + agent routing), Master Data (9 canonical lists seeded — US states, lines of business, perils, occupancy, ISO construction, treaty types, currencies, Saffir-Simpson, decision outcomes), Ontology (entities, relationships, tasks, processes — the domain model behind every agent), and Process Discovery (reads your live data to surface bottlenecks + recommend the next move).
Admin surface · enterprise tier
◆ Dynamic sub-agents
Reflexive mini-agents that spawn when a parent specialist flags something.
Define focused sub-agents the pipeline spawns dynamically: verify a postcode against FEMA, corroborate loss history against ISO ClaimSearch, look up a clause cited in a coverage memo. Five customisable categories — verification, research, synthesis, data fetch, custom. A pure-function reflexive suggester matches trigger patterns (equals · contains · regex · array_includes · truthy) against parent agent JSON and either auto-spawns or surfaces a click-to-spawn chip in the run UI. Every spawn writes a sub_agent_invocations row with parent/child trace lineage.
See sub-agent builder
◑ Built for US P&C carriers
Native to a carrier UW desk, not adapted from an MGA tool.
Multi-state filings, direct-authority bind, NAIC market-conduct posture, treaty cession across CAT XL / quota share / surplus share — every assumption that hardcodes a Lloyd's coverholder workflow has been lifted out and made line-of-business agnostic. US P&C is the default; UK Lloyd's is a configuration.
◆ NAIC + state DOI
Rate filing + market conduct, in one audit pack.
Every memo carries the prompt version, model used, source citations, and the underwriter's rationale — exportable as a state DOI rate-and-form filing review packet, an NAIC market-conduct response, or an internal audit log. The trace log is the single source of truth.
◆ Multi-state P&C
Filed-rate adequacy, by state.
Submission-vs-appetite scoring runs against your filed appetite per state. Technical-vs-filed deviation is flagged before bind so an admitted carrier doesn't accidentally cross a CA Prop 103 prior-approval threshold or trigger a NY informational refiling.
◆ Direct authority
Full carrier UW bind, not coverholder gating.
Bind authority lives with your underwriters. No "exceeds delegated authority" referral hardcoded into the workflow — replaced with your own bind-authority limits per UW, per line, per state. Configurable per role.
◆ Treaty cession
CAT XL, quota share, surplus share — all on file.
Connect your treaty programme (Carrier Treaty Hub data fabric tile, on the roadmap). Each bind shows cession + retention impact pre-bind; CAT-mode generates cedent letters and treaty notification drafts in minutes when an event is declared.
◆ Enterprise deploy
SSO + VPC + BYOLM. Procurement-ready.
SAML / OIDC SSO with SCIM provisioning. VPC or on-premises deployment options for inference traffic to never leave your perimeter. Bring-your-own LLM across Anthropic, OpenAI, Vertex, Bedrock. SOC 2 prep + DPA + vendor due-diligence pack on day one.
◆ Audit-grade memo
Every bind defensible without "let me pull that file".
agent_traces logs every model call (prompt version, tokens, latency, structured output) with submission lineage and the final UW decision. When NAIC or a state examiner asks "why was this risk bound", the chain reconstructs in seconds, not a week.
◑ Submission Journey
From intake to bind in one guided flow
Stage 1 · 0–4s
Submission ingested
Email, PDF, spreadsheet, and broker notes are normalized into a single structured record.
Stage 2 · 4–12s
Core parse and enrich
Entity, exposure, occupancy, and location fields are extracted and cross-checked against data fabric signals.
Stage 3 · 12–25s
Agent fan-out
Risk, flood, treaty, pricing, compliance, and portfolio specialists run in parallel and cite every assertion.
Stage 4 · 25–30s
Memo and bind gates
A structured recommendation memo lands in Command Center with confidence chips and clear next actions.
◑ Go Deeper
Evaluate the platform from strategy to implementation.
Platform capabilities
See the full architecture: Command Center, Pipeline DAG, Agent Studio, data fabric, traces, and governance.
Explore platform
Role-based outcomes
Understand what underwriters, MGAs, brokers, carriers, and reinsurers each gain from a shared system of action.
See solutions
Credits and economics
Transparent per-agent metering, predictable spend bands, and practical deployment models for underwriting teams.
View pricing
Operator notes
Practical writing on shipping agentic underwriting in regulated environments with traceability intact.
Read the blog
◑ Watch
The full pipeline walkthrough in 75 seconds.
Why Vortic
Three things you cannot do with a spreadsheet.
Decisions in 30 seconds
Nine specialist agents fan out in parallel on every submission. Parse, risk, flood, pricing, compliance, treaty, portfolio, memo. You get a structured decision memo before the broker has refreshed their inbox.
Three views, one source of truth
Command Center is your daily view — what needs you today. Workbench is your detailed view — case files, decisions in flight, learning loops. Run Pipeline is your playground — paste a submission, run specialists, watch them stream. Same data underneath.
Compliance built in
Every agent run is traced. Every decision has a memo. Appetite, NAIC (US) / FCA (UK) checks, treaty caps and concentration alerts run on every submission — not as a quarterly audit. Your audit trail writes itself.
Design partners
What underwriters say.
Cut our average decision time from 4 hours to 28 seconds. Underwriters spend their day deciding, not chasing.
Head of Underwriting
US E&S MGA · coastal property
Sample
The audit trail alone paid for the year. Every memo writes itself with sources cited.
Head of Operations
Program administrator · Florida + Texas
Sample
Command Center is the only piece of software my team opens before email.
Active Underwriter
Wholesale broker · Northeast specialty
Sample
Built for regulated environments
SOC 2 Type II
In progress
ISO 27001
In progress
GDPR-ready
Compliant
EU & US regions
Available
◑ FAQ
Questions teams ask before rollout
What problem does Vortic solve for underwriting teams?
Vortic compresses the loop between intake, analysis, and decision. Broker submissions become structured data, specialists run in parallel, and underwriters receive a cited memo in one surface where they can bind, decline, refer, and coordinate tasks.
How quickly can teams expect a full analysis?
Most runs complete in tens of seconds with streaming updates. The exact latency depends on submission complexity and model routing, but every stage is visible so teams understand where time is spent.
How does Vortic support compliance and audit requirements?
Every call is traced with model, prompt lineage, and output. Memo assertions are citation-backed and the final human decision is logged with context, producing an audit trail suitable for regulated review.
Can we adapt it to our own appetite and data sources?
Yes. Agent prompts, model tiers, and data-fabric connectors are configurable. Teams can add internal warehouse feeds and appetite logic while preserving the shared operational surface.
◑ Deep Dive
Underwriting Platform Architecture Built for Regulated Insurance Operations
The platform layer combines an orchestration engine, configurable specialist agents, source-grounded data fabric, and append-only decision trace infrastructure. This gives underwriting teams predictable execution, clear governance posture, and faster production rollout.
- Pipeline DAG orchestration for parse, specialist fan-out, and memo synthesis phases.
- Agent Studio controls for model routing, prompt governance, and source-level grounding rules.
- Data connectors for peril, geospatial, firmographic, policy, and internal warehouse context.
- Audit and compliance instrumentation aligned to insurer model-governance expectations.
Ready to lift underwriting output without losing control?
Join teams using Vortic to move faster on submissions while keeping human judgment and audit posture intact.