AI Decision Intelligence
Cyxhome ingests market, pricing and risk data around the clock, runs it through predictive models, and delivers recommendations with a daily report attached. You review outcomes; you do not chase data.
Built for UK-based investors and business owners managing portfolios or operational decisions that require consistent, evidenced analysis rather than occasional review.
The Shift
Most decision processes still depend on someone opening a spreadsheet, cross-checking figures, and forming a judgement under time pressure. Cyxhome replaces that cycle with a standing model that watches the inputs continuously and only surfaces output when a decision point is reached.
Price, volume, macroeconomic and operational data feeds are pulled and normalised without manual entry or scheduled review.
Patterns are scored against historical and live data to produce a probability-weighted view, not a single static forecast.
Every recommendation carries a stated confidence level and exposure estimate, so you can size a decision before acting on it.
Transparency
A model is only as useful as the ability to check it. Cyxhome publishes a daily report against every open position and every closed recommendation, so performance can be reviewed line by line rather than taken on trust.
Methodology
The process is sequential and repeatable. Nothing is generated without passing through each stage below, which is what allows the output to be audited after the fact.
Step 01
Structured and unstructured data — pricing feeds, filings, operational metrics — is collected continuously and normalised into a common format.
Step 02
The normalised data is run through trained models that estimate likely outcomes across defined time horizons, with confidence intervals attached.
Step 03
Each predicted outcome is weighted against exposure limits and volatility thresholds, filtering out recommendations that fail risk criteria.
Step 04
What remains is issued as a specific recommendation with a stated rationale, which is then tracked in the following day's report.
Applications
The underlying model does not change between sectors; only the data inputs and risk parameters do. Three common applications are set out below.
Operational and market data are analysed together to flag when a planning assumption — pricing, demand, supplier cost — no longer matches observed conditions, prompting a review before the assumption causes a loss.
Positions are continuously checked against volatility and correlation thresholds. Where exposure drifts beyond a set tolerance, the system issues a flagged alert with supporting data rather than a generic warning.
Allocation recommendations are issued when the model identifies a materially better risk-to-return balance, rather than on a fixed rebalancing schedule, reducing unnecessary turnover.
Questions
Data supplied for analysis is processed solely to generate recommendations and reports for your account. Access controls restrict who within Cyxhome can view raw inputs, and retention periods are set to match what is required for audit and report accuracy, not indefinite storage.
Timelines depend on the number and complexity of data sources being connected. A single market data feed can typically be connected and validated within a short onboarding window; multiple internal systems require a longer, staged integration. Documentation covers each connector type.
Every recommendation is scored against its actual outcome and included in the daily report, allowing accuracy to be reviewed over any period you choose rather than relying on a single headline figure. We do not publish backtested projections in place of live performance records.