The Auto Mark Engine.
A bespoke marking and grading engine that delivers instant, rigorous feedback. Calibrated on hundreds of Fellow-marked scripts for precision.
Performance Calibrated.
Verified against qualified actuaries. Designed to catch what others miss.
Human Marker
Standard Calibration
Auto Mark
Engine
Engineered for Rigour.
A bespoke evaluation pipeline, not a generic wrapper.
Vector Logic Validation
We don't just match keywords. The engine validates the directionality of your arguments (e.g., Yields ↑ ⇒ Liabilities ↓). If the vector format is wrong, the mark is withheld.
Verb Calibration
Listing points on a "Discuss" question? The engine knows. Marks are scaled strictly against the command verb's depth requirement.
Fatal Error Shield
A dedicated "Audit Layer" scans for Safety-Critical contradictions (e.g., Prob > 1) that would cause an instant fail in real life.
The Evaluation Protocol
Unlike generic models, Auto Mark follows a strict, multi-stage actuarial audit pipeline to ensure every mark is earned, not given.
A Heritage of Continuous Refinement.
Built on years of actuarial research, calibration, and continuous refinement.
The Inception
First version presented at the Actuarial Convention. A foundational step in automating actuarial feedback.
QA Engine Implementation
We implemented a rigorous QA engine to track actuary marking performance. This gathered critical data on human inconsistencies to train our models.
Auto Mark Live
The marking engine is now live, leveraging insights from thousands of calibrated data points to deliver instant, consistent feedback.
Engine Availability.
Auto Mark is trained and calibrated for the following Core and Fellowship Principles.
More subjects coming soon based on demand.