SYSTEM ARCHITECTURE

Built for evidence-backed portfolio decisions.

A layered intelligence architecture that keeps measured evidence, AI interpretation, portfolio context and capital constraints clear and inspectable.

SYSTEM DESIGN

Evidence first. AI where it adds judgement.

Prospect is not an LLM wrapper around portfolio data. The system uses the best method for each problem: statistics, time-series methods, change detection, machine learning, optimisation, explicit rules and frontier AI models where reasoning or synthesis adds value.

Measured evidence and AI interpretation remain distinct so an investment team can see what the system observed, what it inferred and why.

INTELLIGENCE STACK

Five connected layers, one decision loop.

05

Capital-allocation intelligence

Constraint-aware scenarios, portfolio effects, reasons, uncertainty and decision records.

04

Portfolio intelligence

Correlation, concentration, common exposure, diversification and simultaneous degradation.

03

Regime intelligence

Versioned market context, strategy × regime compatibility and dependency changes.

02

Strategy intelligence

Behavioural fingerprints, strategy-specific monitors, change detection and significance.

01

Strategy understanding

Structured identity, thesis, expected edge, dependencies, assumptions and lineage.

TRUST & PROVENANCE

Institutional intelligence has to be inspectable.

DATA

Traceable evidence

Measured claims trace back to source data, transformations, feature windows, configuration and model versions.

AI

Grounded interpretation

AI explanation is grounded in saved evidence and separated from supplied facts and deterministic measurements.

CONTROL

Explicit constraints

Eligibility, concentration, exposure, drawdown, capacity and approval rules remain first-class controls.

CURRENT BOUNDARY

Human-controlled capital.

Prospect is currently being built as institutional analytics and decision-support software. It can produce evidence, interpretation and allocation scenarios; the customer retains discretion over capital.

The architecture can support deeper automation later if product trust, validation, regulation and customer demand justify it.

BUILDING V1

The MVP is the first executable version of the full company.

It is being built toward strategy understanding, regime intelligence, portfolio intelligence and explainable allocation support—not a standalone monitoring dashboard.

View the product →