Risk Shell AI
Risk Shell AI is a natural-language interaction and assistance layer integrated with the Risk Shell investment management platform. It allows users to work with Risk Shell functionality, analytical concepts and portfolio workflows using ordinary financial language while retaining the platform's established graphical interface and quantitative engines.
The AI layer does not replace Risk Shell's underlying analytical models. Portfolio optimization, factor analysis, stress testing, peer-group analysis and other calculations continue to be performed by the corresponding Risk Shell engines. Risk Shell AI provides the interface between the user, the platform's functionality and the extensive body of product and investment-management knowledge accumulated through more than two decades of development.
Natural-Language Access To Risk Shell
Users can describe an analytical task, ask how a particular Risk Shell function should be configured, or request guidance on an investment problem without having to locate the relevant menu, manual page or training material first. The AI interprets the request in the context of Risk Shell terminology and workflows and provides guidance appropriate to the platform rather than generic financial-software instructions. A query is translated into a screening rule, a factor model, a stress test, a peer comparison, a FlexiRank specification or an optimization, and run on the same data and the same models as the GUI. It is not a general finance chatbot attached to a portfolio system.
- Ask questions about Risk Shell functions and analytical procedures using natural language.
- Receive context-specific guidance while working within the Risk Shell GUI.
- Understand portfolio construction, optimization, factor analysis, stress testing and other quantitative workflows without searching separate manuals.
- Obtain explanations of Risk Shell statistics, analytical settings and model outputs in an investment-management context.
- Move from an investment question to the relevant Risk Shell workflow more directly.
In-House Financial LLM Trained On 2.3 Billion Tokens
Risk Shell AI is based on an in-house language model trained on approximately 2.3 billion tokens of financial, quantitative and Risk Shell-specific material. The training corpus was constructed around the terminology and workflows used by investment professionals rather than around general-purpose conversational content.
A significant part of the training material was derived from ABC Quant's own product knowledge base, including thousands of recorded Risk Shell training sessions, tutorials and instructional videos accumulated over many years. This gives the model detailed knowledge of both the platform interface and the way individual analytical components are used in practice. The assistant therefore uses the product’s own objects — trailing and closed windows, factor shocks, stress-database estimates, importance weights, differential-evolution constraints — rather than a generic description of them.
- Risk Shell product documentation and technical manuals.
- Thousands of recorded training and tutorial videos.
- Investment analytics and quantitative finance material.
- Portfolio construction and risk-management workflows.
- Risk Shell terminology, component relationships and operating procedures.
