The captive scorecard built for you.

The feasibility study, made readable. Then a live scorecard across every placement in the captive, so the parent always knows where the retained risk sits.

reinsurance · excess of retention fronting carrier · specific retention · the waterline retained in the captive · what Keel watches expected retained · 100% policy months → cumulative retained claims
the keel: the weight below the waterline that keeps the structure upright. that's the retained layer, and where we got the name.
● One scorecard, every placement Feasibility study in ten minutes, not ninety pages 0 models trained on your data Sibling to Corridor, sharing one record
Why Keel

The study gets read once. The captive runs for years.

A captive starts with a feasibility study: dense, actuarial, and delivered as a PDF the parent's CFO reads the night before the board vote. Then the captive goes live and the same parent gets a quarterly pack, a renewal deck, and a call from the captive manager when something drifts.

Keel closes that gap. The study becomes a decision the parent can actually read. Then the same numbers keep running as a live scorecard: every placement in the captive, retained against expected, cell by cell, updated as claims come in. The captive manager stops rebuilding the picture every quarter, and the parent stops learning about problems at renewal.

Two halves, one record

Decide it. Then watch it.

01 · Feasibility

The study, made readable.

Keel takes the actuarial study your captive manager already produces and presents it to the parent as a decision, not a document.

  • Retention options side by side: premium saved, capital required, and worst-year exposure for each, on one screen.
  • Assumptions surfaced: loss picks, trend, and credibility called out, with the sensitivity of each.
  • Board-ready summary: the questions a CFO asks, answered in order, with the full study one click away.
  • Re-run on request: change a retention or a line, and the whole picture updates.
02 · Scorecard

Every placement, synchronized.

Once the captive is live, the study's assumptions become the baseline and Keel tracks reality against them across every line and every cell.

  • Retained vs. expected per placement, per cell, per program year, with the drift flagged before it reaches the aggregate.
  • Layer view: what sits with the captive, the fronting carrier, and the reinsurer, and how close each layer is to its limit.
  • Capital and collateral against actual retained losses, so the letter of credit conversation happens early.
  • One pack: the parent, the captive manager, the fronting carrier, and the reinsurer read the same page.
PlacementRetentionExpectedRetained YTDvs. expectedStatus
Medical stop losscell A · PY 2026
$250k spec
$4.10M
$3.62M
88%
On track
Hospital professionalcell B · PY 2026
$2M / occ
$6.80M
$7.35M
108%
Watch
Workers' compensationcell C · PY 2026
$500k / occ
$2.25M
$1.94M
86%
On track
General liabilitycell D · PY 2026
$1M / occ
$1.10M
$1.41M
128%
Aggregate at risk
illustrative · the real scorecard is built to your cell structure, your lines, and your reporting calendar
Who it's for

Everyone who has to explain the captive to someone else.

Captive managers

Sell the study, keep the client

Turn the feasibility engagement into the start of a monitoring relationship. One Keel deployment serves every captive you manage, on your brand.

Parents and boards

Know before renewal

See the retained layer the way you see the balance sheet: current, comparable, and without waiting for the quarterly pack.

Fronting carriers and reinsurers

Cedent quality, visible

The same scorecard the parent sees, scoped to your layer. Fewer surprises at the treaty renewal, and a cleaner conversation about collateral.

✦ Keel AI

An analyst who has read the whole study. And every quarter since.

Ask the captive a question in plain language and get an answer with the arithmetic shown. Every response is logged, and nothing acts without a person approving it.

why is general liability running hot?
Two claims over $400k in cell D, both Q2. Ex those two, cell D is at 91% of expected. This is a large-loss question, not a frequency question.
what happens to capital if we raise the med stop loss retention to $350k?
Expected retained rises $610k; worst-year exposure rises $1.4M. Required capital moves from $5.2M to $6.1M at your current funding ratio.
are we going to hit the aggregate this year?
At current run rate, cell D reaches its aggregate in month 11. Cells A through C do not. Reinsurer notice under the treaty is due at 100%.
draft the board summary for Q3.
Done. One page, four placements, two items flagged. Review it before it goes anywhere.
✦ Keel AI runs inside your environment and learns from your captive's own history. No outside model is trained on your data, and every workflow works with the AI switched off.
Built the Healthmark AI way

The same terms as Corridor.

Built to your structure

Your cells, your lines, your fronting and reinsurance arrangements. Keel is assembled to how your captive actually works, not a template everyone gets.

Own it, or let us run it

Build-and-handover puts the code in your walls. The managed subscription carries hosting and updates for you. Either way, the data stays yours.

Audit, everywhere

Every assumption change, every re-run, every AI suggestion and who accepted it. Logged as it happens, so the board pack can always be traced back.

Two ways to work with us

Own it outright, or let us run it for you.

Either way the build is to your specifications and nothing trains our models. The difference is ownership: take the keys, or let us carry it.

Build & handover

One-time engagement

We build Keel to your captive's structure, deploy it inside your environment, and hand over code and documentation. Your team runs it from there. Support available on request.

Managed subscription

Recurring engagement

We build to your structure, then run it for you: hosting, monitoring, new cells and lines as the captive grows, and model upgrades. The platform is licensed, not owned; your data and assumptions remain yours. Convert to full ownership at any time.

See it

Bring a feasibility study.

Thirty minutes, your team, and one real study. We'll show you what the parent would have seen, and what the scorecard looks like a year in.

To set one up
Email Simon at simon@healthmark.ai