Oriol Cervantes · Decision Intelligence

01

The experience behind the models.

Before any engine on this page, the same discipline ran by hand, at institutional scale, for years. Three pieces of that work, with what each one helped decide and how far it can be checked.

Heatmap of 150 countries by 200 categories, redrawn: a few hundred pairs carry most of the spend — ~2017 · redrawn from the Excel original · shape, not figures

Spend mapping · UNDP · 2011 to 2021

One heatmap, 150 countries by 200 categories.

Every year, about 500,000 purchase-order rows of sixty fields each (some 30 million cells) were cleaned, reconstructed and summarised into 30,000 cells on a single map. Sorting by value concentration showed that a small fraction of country-category pairs carried nearly all institutional spend.

What it helped decide Where procurement strategy should concentrate, and which offices needed support first. The office clustering later fed the agency's 2018 business case for global shared services.

USD 11Ma year in savings potential, identified by the 2017 clustering model

My contribution Designed and ran the method, the models, the reports and the training. The USD 11M is the model's 2017 estimate, not an audited total.

Professional experience · first-handBenefit quantification deck, 2017, available for discussion where permitted

Supply positioning matrix, expenditure by business risk, 39 categories, redrawn — ~2018 · redrawn from the Excel original · 39 categories, positions illustrative

Supply strategy · UNICEF Supply Division · 2007 to 2010

Dashboards adopted across 145 countries and 27 categories.

A category-management analytics layer over global procurement, integrating ERP data to support supplier, product, client and country strategies, with an automated dashboard methodology that country teams could run themselves.

What it helped decide Which categories to consolidate, which suppliers to position where, and where efficiency gains were available.

145countries running the same dashboards, 27 categories

My contribution Designed the methodology and the dashboards, and trained the country teams that ran them.

Professional experience · first-handSupporting material available for discussion where permitted

Benchmarking · WHO Regional Office for Europe · 2019 to 2020

A standard score across 53 country offices.

A scoring and benchmarking framework for performance trends and data maturity, so offices could be compared on the same basis during organisational change, with the quality of each office's data scored and shown, not hidden.

What it helped decide Which offices needed targeted support, and which comparisons could be trusted.

3components per office: performance trend, data maturity, data quality

My contribution Led the design and rollout; wrote the scoring method and the visual standard.

Professional experience · first-handFramework sample available for discussion

Roles and dates in full on LinkedIn and in the CV. Figures come from professional experience and are labelled by how far they can be checked.

02

Three decisions, rebuilt as models you can play with.

Each one started as a real kind of decision. The problem, my contribution, and what you can test yourself. These are personal prototypes: they show the method, they do not report results achieved for any client.

150 country offices · USD 3.25B · 160,550 purchase orders · synthetic data

Institutional procurement personal prototype · synthetic demonstration data

Paretum

Where this comes from. Ten years of spend mapping and office clustering at UNDP, rebuilt as an agency-neutral engine on synthetic data.

The problem

A network of 150 offices orders USD 3.25B a year through 160,550 purchase orders. Standard reporting shows the transactions. It does not say where to look first, or which number can be trusted.

My contribution

The spend-mapping and office-clustering approach from the first piece of experience above, rebuilt with AI as an agency-neutral engine: a five-mark scorecard (fit, efficiency, price, resilience, local economy) against 150 peers, with every finding opening its chart, its table and its method, in that order. Everything is a signal to review, never a verdict; the controller decides.

you are the health minister · set the budget · watch the system respond

Public health & financing personal prototype · public data and illustrative scenarios

Health Systems Are Decision Systems

Where this comes from. The scoring and benchmarking work across 53 WHO country offices: the same question, which comparisons can be trusted, applied to a whole health system.

The problem

Where a child is born sets a thirty-year gap in life expectancy before a single choice is made. Global health has no shortage of actors or money; they flow in different directions, and the trade-offs between them are rarely visible to the people allocating the budget.

My contribution

Public-health, financing and outcome data reconciled into one model, then challenged by four AI models for gaps and overclaiming before any chart was drawn. Built as six scroll-driven chapters where the reader holds a real-sized budget and moves the money, and the ecosystem of funders, ministries and communities responds.

4 assumptions · 100 scenarios · holds or breaks

Decisions under uncertainty personal prototype · worked example

The Fragility Lab

Where this comes from. Years of watching strong recommendations fail on one untested assumption. The lab makes that assumption visible before anything is committed.

The problem

A 4.8× recommendation looks robust on paper. Then reality intervenes, and the biggest number is usually the one whose model is most wrong in your favour.

My contribution

Each of four parameters is estimated from imperfect evidence. The tool runs a hundred scenarios and maps the whole parameter landscape, so the gap between the optimistic number and the realistic spread is visible before anything is committed. Knowing where a model breaks in the field comes from twenty years of institutional work, not from the statistics.

05

More from the shelf.

The same method, on subjects people love. Public data, every figure labelled, and the reader in control of the criteria.

The Civilisation Lab194 nations, 100+ indicators, 1900 to 2050

Last Mile Lab120,000 vaccine doses, only 60 in 100 reach a child

Atles dels Castells1,453 castles, a thousand years, year by year

Essay · The Three Broken Laws of Developmentcarbon, fertility and inequality: three laws that turned out to be conditionalread ↗ Carta Blanca Labsgames, atlases, languages and what is still in the ovenbrowse ↗ Studio & the generative archivevisual direction, Prism, and a long generative practice that predates the current AI wavein the Labs ↗

Hard decisions, turned into models you can interrogate.

The experience behind the models.

One heatmap, 150 countries by 200 categories.

Dashboards adopted across 145 countries and 27 categories.

A standard score across 53 country offices.

Three decisions, rebuilt as models you can play with.

Paretum

Health Systems Are Decision Systems

The Fragility Lab

One promise, verified twice.

A council of models, one judgement

The foundation first

Not one model. A council I direct.

One of the engines, running.

About me.

Let's talk. About decisions and evidence.

Thinking about decision intelligence, data governance or AI in an institution?

Working on public data, a teaching case or a talk?

More from the shelf.

Disclaimers

Sources