Technology and AI
Intelligence that earns its place
We use artificial intelligence where it genuinely improves a decision, and we say plainly where we do not use it at all. Several tools on this site deliberately contain no AI, because fixed rules give you an answer you can audit.
Our position
AI as a capability, not a label
Plenty of software claims intelligence it does not have. Our rule is simple: if a result can be produced by transparent rules, we use rules and tell you so. We reach for a model only where the problem genuinely needs one, and we show our working either way.
The self-assessment on this site is a worked example. It scores your answers with fixed, published weights. The same answers always produce the same result, you can see why, and no model is involved anywhere in it.
That is not a limitation we are apologising for. For a scoring tool, auditability beats sophistication. Where prediction genuinely helps, such as forecasting or anomaly detection, we use models and label them clearly.
- Rule-based scoring where the logic should be inspectable
- Predictive models where patterns genuinely beat rules
- Every automated recommendation traceable to its inputs
- No model is ever the final word on a business decision
- Clear labels on which tools use AI and which do not
Capabilities
Where intelligence is applied
- Rule-based scoring
Transparent, published weights across the five lenses. Identical inputs give identical results, every time, and the calculation is documented.
- Predictive analytics
Trend detection and forecasting on commercial data, with confidence stated rather than implied.
- Anomaly alerts
Automated monitoring that surfaces the figure that moved unexpectedly, instead of waiting for someone to notice at month end.
- Customer modelling
Behavioural analysis to identify which customers resemble your most valuable ones, and which are drifting.
- Natural language queries
Ask a question of your own business data in plain English or Persian, and get the figure with its source.
- Scenario modelling
What-if analysis on price, channel and volume so a decision is tested before it is committed.
Natural language queries and scenario modelling against your own connected data are in development. The scenario simulator in the experience centre runs today on sample data.
Engineering
How the platform is built
- Architecture
- Business logic is kept separate from the platform it runs on, so the rules can be tested on their own and audited without a running website.
- Testing
- The scoring engine is verified against several thousand generated cases, and the browser and server implementations are checked against each other so they can never disagree.
- Authorisation
- Access is denied by default and granted explicitly per record. A request for something you may not see is indistinguishable from a request for something that does not exist.
- Audit trail
- Portal actions are written to an append-only, hash-chained log, including refused attempts. Records cannot be altered after the fact without detection.
- Money handling
- All amounts are stored as whole minor units with an explicit currency code, so rounding errors cannot accumulate and no figure is ever ambiguous.
- Privacy
- The self-assessment never transmits your answers. The enquiry form stores nothing on the server and forwards your message by email.
Get started
Want the technical detail?
We are happy to walk your technical team through the architecture, the authorisation model and the audit design before you commit to anything.
- No obligation
- Confidential
- Tailored to your industry