Home
About us
Solutions
Case studies
News
Intelligence
PriceSense
Reading where prices are going
from the signals the
market already gives.
Raw-material price analysis · generative AI and machine learning · human-in-the-loop
Schematic representation: PriceSense produces a directional estimate, not a quotation and not a trading signal.
The context
The signal is there.
It just arrives
late.
In raw-material markets, price moves are usually preceded by news and events surfacing from different sources. Collecting that information, judging what matters and tying it to the historical data takes time. By the time the move shows up in the price lists, part of the useful information has often been available for days.
Volatility eats into margin
Price swings change procurement costs and margin. Judging a negotiation means looking also at the signals that come before the price lists move.
The information comes from everywhere
News, reports, trade publications and market events make up a wide picture. Finding what is relevant and connecting it to the categories you actually buy is continuous work.
Historical data is not enough
Historical series describe where prices have been, but on their own they do not explain what may move them next. Reading the market means putting context alongside the numbers.
One expert, several markets
The buyer’s experience is what makes the signals readable, but following several categories, markets and sources every day is a lot. A tool can organise the information and put it in front of the expert.
How it works
From market signals
to a price estimate
the expert has validated.
PriceSense collects news, events and market reports, and relates them to the historical series and the economic context. The models produce a directional estimate, which goes to the expert for review before it supports any decision.
Human-in-the-loop
The technology produces the estimate.
The expert decides.
The estimate is never used automatically. The expert who knows the category and the market reviews it, looks at the signals behind it, and decides whether to use it, dig further or discard it. The technology supports the analysis; the final judgement stays with the person doing the buying.
The output is a directional estimate — up, down or flat — together with the information that produced it, which the expert can read while reviewing.
Expert validation is part of the process: the estimate is reviewed before it supports any decision.
What it brings together
Historical data and signals
to read the market.
PriceSense combines three kinds of information. The historical series give the quantitative reference, the signals help interpret what might move the price, and the economic and market context frames the whole reading.
Historical price series for the raw materials being followed
News, events and reports that can move the relevant markets
Economic and market context in which to read the data and the signals
Inside the product
One estimate,
three places it is used.
The directional estimate supports different steps along the procurement process: judging buying conditions, planning costs, and watching margin.
Judging when to buy
Whoever runs procurement can read the expected direction and the signals behind it, and bring that into the assessment of negotiations and supply terms.
- Directional estimate for the categories followed
- The signals attached to each estimate
- History of the estimates, compared with what actually happened
Supporting the plan
The estimates can feed supply planning and cost assumptions, alongside internal data and whatever else the budgeting process uses.
- Estimates supporting the cost assumptions
- Clear view of the categories being followed
- Periods and markets compared
Watching the effect on margin
The expected direction adds one more element for judging how purchase costs might move and what that does to margin — without replacing financial analysis or management judgement.
- A summary by category and market
- Signals of a possible price move
- Expert validation kept on record
The scope
What the estimate does and does not say.
| Aspect | PriceSense | It is not |
|---|---|---|
| Output | A directional estimate: up, down or flat | A quotation, or a guaranteed future price |
| Use | Support for buying decisions and planning | An automatic instruction to place a trade |
| Method | Historical series, market signals and context analysed with generative AI and machine learning | A black box: the signals used stay open to inspection |
| Governance | Expert validation before the estimate is used | A system that decides instead of the buyer |
Keep exploring
The other Add Value products
Data · Cloud and AI
frAIme
Integrates different data sources into a cloud-native data fabric, leaving the information in its systems of record and making it available to AI applications.
Intelligence · Mainframe
OptiMIPS
Analyses mainframe consumption and attributes it to the applications, to find where resources can be optimised.
Intelligence · Procurement
TSV
Categorises indirect spend with AI and makes it searchable by category and supplier.
Intelligence · Insurance
Calamìta
Cross-references seismic, weather and hydrogeological data with geolocated policies to analyse the NatCat exposure of an insurance portfolio.
Intelligence · Banking
Tequery
Lets you query databases in plain language and check the SQL generated behind every answer.
Software · Modernisation
L2M
Core applications, modernised and verified against real data.
Intelligence · AI adoption
Officina AI
Finding where artificial intelligence can create value in business processes.