Case study · Pharma · Logistics

Predictive control tower for the pharmaceutical supply chain

Unify the whole chain in a single place. Anticipate supply tension. Decide on data, not on intuition.

Unified data platform

Predictive demand models

Early risk alerts

Where it all started

A critical chain managed without a single view

A pharmaceutical company with an international supply chain lived with a structural weak point: it had no single view of its own procurement.

Supplier, production, inventory and demand data lived in systems that did not talk to each other.

When supply tension appeared (a delay from an active ingredient (API) supplier, a demand spike), the team found out late, with stock already committed and little room to react.

The challenge

Procurement, inventory and demand information spread across systems that did not talk to each other.

Heavy reliance on single-source active ingredient (API) suppliers, with no alerts when a lead time stretches.

Reactive planning: tension was spotted once it was already a stockout (OOS).

Batch and expiry traceability (FEFO) hard to consolidate for responding to audits.

Purchasing decisions based on individual experience, not on consolidated data.

What we proposed

Unify the whole chain into a single data platform, procurement, production, inventory, distribution and demand, and build on top of it predictive models for demand (with measured forecast accuracy) and for supply risk. All in a Power BI with early alerts that flag when an indicator enters a risk zone. The foundation for an S&OP that looks at what is coming, not at what already happened.

The shift

From putting out fires to getting ahead of them

Before

Chain data scattered across isolated systems

Supply tension spotted once it was already a stockout

Reliance on suppliers with no early warning signals

Planning based on intuition and spreadsheets

Now

A single source of truth with visibility across the whole chain

Predictive models that anticipate tension before it arrives

Automatic alerts when an indicator enters a risk zone

Purchasing and planning decisions backed by data

links unified

predictive models (demand and risk)

control tower in Power BI

The value generated

What changes when the chain becomes predictable

End-to-end visibility

The whole chain in a single dashboard, from supplier to point of dispensing.

Anticipation, not reaction

The models detect supply tension while there is still room to act and avoid the OOS.

Alerts that arrive on their own

The system warns the right team, with no manual watching, when an indicator enters risk.

Less hidden dependency

The risk of concentration on a single API supplier stops being a blind spot (lead time and delays under control).

Traceability under control

Batches, expiry dates (FEFO) and movements consolidated and available for audit, in line with GDP.

Service level and S&OP

Better OTIF / fill rate and an S&OP fed with demand and risk, not with intuition.

What the client says

What used to be detected late and solved through emergencies is now anticipated with time to spare. The supply chain team plans and buys looking at what is coming, not at what has already happened, with the ability to get ahead and without giving up the control a regulated environment demands.

Supply Chain Management

Pharmaceutical company

F&Q

Frequently asked questions

What is a supply chain control tower?

It is a single command center that brings together data from the whole chain (procurement, inventory, distribution and demand) to see, anticipate and decide from one place, instead of chasing information across several systems. At BertIA, an AI and data consultancy based in Barcelona, we apply it in pharmaceutical settings where supply is critical and every decision counts.

How is it different from a traditional dashboard?

A traditional dashboard shows what has already happened; a predictive control tower adds models that anticipate what is going to happen. You do not just see today’s stock and demand: the system estimates the supply tension that is coming and triggers early alerts when an indicator enters a risk zone, with time to act before the stockout (OOS).

Are predictive models reliable for purchasing decisions?

The models support the decision, they do not replace it. They learn from demand history, from each supplier’s behavior (lead time, recurring delays) and from inventory levels to estimate risk, with measured forecast accuracy. The person keeps the purchasing and planning decision; the model provides the early signal and the judgment backed by data.

How does it help against an active ingredient shortage?

By integrating each supplier’s behavior and the level of dependency per reference, the system detects early risk signals (recurring delays, concentration on a single source) and alerts before the tension turns into a stockout.

What does it take to implement it on an existing data platform?

If a data platform already exists, at BertIA we reuse it as the core and add the model layers and the control tower on top. That lowers the cost and time of implementation and avoids duplicating infrastructure, supporting adoption so it becomes part of daily operations.