VorticVortic
◐ Platform

Nine specialist agents.
One source of truth.
Every claim cited.

Vortic is the AI underwriting platform built for MGAs, brokers, Lloyd’s coverholders, and US E&S underwriters who need bind-ready decisions in seconds — and an audit trail their regulator will accept.

Start with $5 free Talk to the team →

No credit card required to start · Pay-as-you-go after the free credit

◐ The gap

Underwriters don’t need another chat window.

Generic AI tools fail two ways: they hallucinate flood zones because the training data is stale, and they leave no audit trail your regulator will accept.

Vortic is built for the specific shape of the underwriter’s day — the lookups, the citations, the team coordination, the bind gate, the regulator. Not as a chat assistant grafted onto insurance.

◑ The product

Three views. One source of truth.

Different cognitive density for different jobs. The data is shared; the surface adapts to what the underwriter is doing right now.

Daily

Command Center

What needs your attention today.

risk
flood
memo

Pre-surfaced: SLA breaches, pending decisions, yesterday's closures, concentration alerts. A chat that already knows your day. Click through to bind, decline, refer, or delegate.

  • ·Contextual chip suggestions seeded from your data
  • ·Three-envelope chat (answer / navigate / task)
  • ·Per-agent credits ledger visible inline
  • ·First thing you see when you sign in
Detailed

Workbench

Deep canvas for one submission.

Every signal we have on one risk: parse output, agent verdicts, similar bound risks, treaty utilisation, portfolio impact. The surface where the underwriter spends their judgment minutes.

  • ·Side-by-side broker PDF + extracted fields
  • ·Agent-by-agent reasoning panel
  • ·Citation links to every source
  • ·One-click prompt override per agent
Playground

Run Pipeline

Try a hypothesis. Now.

Drag a PDF, hit run, watch nine agents stream their analysis live. The space to test prompts, try new agents, replay runs from history. No queue, no consequence — just experiments.

  • ·Streaming SSE timeline of all 9 agents
  • ·Replay any historical run
  • ·Per-run prompt overrides without saving
  • ·Cost preview before fan-out
◑ Why this works

Built for the moment of regulatory scrutiny.

Verifiable, not hallucinated

Every flood zone, postcode, sanctions check, and regulatory threshold is sourced from a verified API and cited in the memo. The model never makes up the facts. If the lookup fails, the memo says "no data" instead of guessing.

Audit trail by construction

Every model call logged with prompt version, tokens, latency, output. Every memo cites EA / FEMA / postcodes.io / your loss-history. Every decision captures the underwriter, the rationale, the agent outputs at that moment. Built for Lloyd's coverholder oversight + PRA SS3/19 from day one.

Cost transparency

Every LLM call records what we paid OpenRouter and what we charged you. Visible in the admin panel as margin per agent. No opaque "AI tax" — when our cost goes up, you see it.

Data stays where it should

PII redaction before any model sees the submission. Per-agent control over which ground-truth lookups join the prompt. Enterprise customers can route to their own private LLM deployment so no inference traffic leaves the VPC.

◐ Voices

What underwriters say.

01 / 11

The audit trail was the unblocker. Our compliance team agreed in one call — they could see every model call, every prompt version, every source citation. That doesn't happen with general-purpose AI tools.

Maya, illustrative
Head of Underwriting
UK MGA, commercial property up to £10M TIV
Each quote represents the composite feedback from early underwriter conversations and is attributed pseudonymously by role and segment. Once design partners consent to be named, real attributions replace these. Want to be quoted? Become a design partner →
◑ Honest comparison

How Vortic actually compares.

We don’t fabricate competitor weaknesses. Where Federato or Cytora or rolling-your-own actually does something better, we say so. Buy the platform that fits your shape, not the one with the cleverest comparison page.

CapabilityVorticChatGPT / Generic AIFederato / CytoraBuild it yourself
Multi-agent orchestration
Specialist agents (parse, risk, flood, pricing, compliance, treaty, portfolio, memo) running in parallel — not one big prompt.
Verified ground-truth lookups
EA Flood + postcodes.io + FEMA NFIP cited inline in the memo. Never made up.
Lloyd's / PRA-grade audit trail
Every model call logged. Every memo cites its sources. Decision history immutable.
Pay-as-you-go pricing
$0.10/credit, $20 minimum, no contract.
Time to first live submission
Two weeks for a typical MGA design partner.
Bring your own LLM (Enterprise)
Anthropic, OpenAI, Google Vertex, AWS Bedrock, or self-hosted.
Human-in-the-loop bind gate
No auto-bind. Underwriter clicks; system traces.
Webhook ingest + dispatch
Bearer-key inbound submissions, HMAC-signed outbound events.
Prompt versioning + rollback
Edit any agent prompt; one-click rollback to a prior version.
Cost-transparency admin panel
See your real OpenRouter spend per agent, per call. No opaque AI-tax.
Big-team enterprise contracts
Dedicated CSM, custom SLAs, on-prem deployment, SSO with SCIM.
Mature actuarial pricing engine
Akur8-style GLM modelling for portfolio rate calibration.
Yes Partial / depends on tier Not supportedComparison data based on each vendor's public materials as of 2026. We update when their positioning changes.
◐ Data fabric

Verified ground truth, cited inline.

Add a data source →
SourceTierWhat we use it for
EA FloodLiveUK flood zone + surface water risk per postcode
postcodes.ioLiveUK postcode → lat/lng + administrative geography
OpenFEMA NFIPLiveUS flood zone per ZIP via official FIRM panels
OpenStreetMap / NominatimLiveReverse geocoding for site-level enrichment
Open-MeteoLiveHistorical weather signals for cat exposure
USGS Earthquake APILiveUS seismic exposure context
Verisk ISOEnterpriseProperty characteristics, BCEGS, ISO loss costs
D&B FirmographicsEnterpriseInsured entity verification + parent/subsidiary
Lloyd's Market DataEnterpriseCoverholder + syndicate market intelligence
Your data warehouseConfigurableSnowflake / Postgres / S3 — your bound risks, loss runs, treaty programme
◑ Stack

Built on infrastructure underwriters can audit.

FrontendNext.js 16 App Router · Tailwind v4 · Framer Motion
BackendSupabase Postgres + Row Level Security · Edge functions
AuthSupabase Auth (email + OAuth) · Cloudflare Turnstile · 2FA-ready
AI gatewayOpenRouter (multi-model routing) · Anthropic / OpenAI / Vertex / Bedrock for Enterprise
StreamingServer-Sent Events for live agent token streams
EmailResend (transactional + invites + alerts)
BillingStripe Checkout · idempotent webhook · per-agent metering
Rate limitUpstash Redis (fail-closed in production)
Auditagent_traces + llm_usage_events + submission_decisions tables
DeployVercel · daily cron · region pinning · Speed Insights
Design partner programme

First three MGAs get pricing locked.

$0.05/credit for 12 months, direct line to the founder, weekly product calls until your specific workflow is solved, co-development of bespoke agents (construction class scoring, marine cargo, your line).

Become a design partner Try with $5 free credit →