Orbotrice Applied AI
Reserving Review Agent

Sign in

This prototype is access-controlled. Select your name and enter your password to continue.

That password was not recognised for the selected name.
For demonstration use. This is a basic access gate, not a security control — do not place confidential data behind it.
Orbotrice Applied AI
Reserving Review Agent · check a reserving submission
Ready · audit on

The submission to review

Upload a reserving submission — your own work to check before submitting, or a submission you are reviewing. The agent checks the selected factors, tail, and reserve against what the data supports.
📄
Drop the submission file
CSV · click or drag
— or —
Triangle–
Selected factors–
Tail factor–
Proposed reserve–

Fill any gaps the file did not contain

The reserve the submission proposes.
The tail factor the submission selected.
A benchmark tail from your own experience study, if you have one. Left blank, no benchmark is claimed.
A margin you hold above the best estimate (IFRS 17 risk-adjustment style).
A confidence level you hold the reserve to. You may enter a margin, a percentile, or both — if both, the tool reconciles them.
The CoV of the reserve, if you have one from a Mack calculation or experience study. Used to fit a lognormal distribution. Left blank, the tool estimates it from the spread between the two methods.
Total earned premium for the triangle. Enables the Bornhuetter-Ferguson and expected-loss-ratio cross-checks. A per-year “Earned Premium” row in the CSV is used if present. Left blank, only chain-ladder runs.
Expected ultimate loss ratio. Used by the Bornhuetter-Ferguson and expected-loss-ratio methods — it only affects the result when earned premium is supplied.
The agent recomputes from the data and compares against the selections in the submission. Every figure is derived from the uploaded submission or the data within it. Triangle dimensions are flexible — any number of accident years and development periods. For demonstration use.
🧮

No submission loaded

Upload a reserving submission. The agent will set the selected factors and tail beside what the data supports, recompute the reserve independently, and attribute any gap to the specific assumptions that caused it.

The submission CSV can contain the triangle alone, or the triangle plus the selected factors, tail, and reserve. Recognised row labels:

2019,1180,1842,2210, ...  ← triangle rows
2020,1310,2074,2502, ...
...
Selected LDF,1.585,1.175,1.082, ...  ← selected factors
Tail,1.012  ← selected tail
Earned Premium,3756000,4170000, ...  ← per year (enables B-F & ELR)
Proposed Reserve,4600000  ← proposed reserve
The submission’s selections checked against what the data supports
Orbotrice Institute
Working prototype · Orbotrice Applied AI · figures derived live from the uploaded submission · applied.orbotrice.com