The Story · NDA

YPF

Turning fragmented requirements and inherited work into a buildable self-service MVP — then de-risking the interface direction without blocking engineering.

Explore YPF’s Future Station
ROLE
Sole Product Designer
SCOPE
MVP Definition
YEAR
2018–2020
VIA
Edrans
YPF self-service kiosk in use at a fueling station
A proposed interface direction selected through comparative validation, and later deployed in YPF stations.
BRIEF

Create delivery certainty before adding more design.

Problem — Detailed requirements, inherited partial work, unresolved flows, and multiple technical teams were converging on a fixed delivery window.

Decision — Define the MVP boundaries first, translate operational requirements into buildable system behavior, and separate product uncertainty from interface work.

Result — Engineering gained a credible path to delivery, while YPF could evaluate a stronger interface direction without putting the committed scope at risk.

FROM AMBIGUITY TO SCOPE

I treated the kiosk as one touchpoint in a larger operational system.

The customer experience depended on more than the screen: fueling hardware, payment, station personnel, cloud services, and teams across Software Development, SiteOps, CloudOps, and AWS LATAM all affected the same journey.

With my frontend background, I worked with Engineering to determine what the MVP could realistically support and made the boundaries explicit:

  • what had to ship,
  • what could be postponed,
  • which actor owned each state,
  • and which technical dependencies had to be resolved before implementation.

The objective was not to shrink the product. It was to remove uncertainty early enough that design and engineering could make reliable decisions.

  • Diagram of the self-service system connecting customer, kiosk, station staff, and cloud services
    Actors and system boundaries
  • End-to-end flow of the self-service fueling process
    End-to-end fueling flow
  • The kiosk touchpoint within the station environment
    The kiosk inside station operations
  • Operational view of a station during self-service fueling
    A station mid-service
  • Mapping fragmented requirements into a scoped MVP plan
    Fragmented inputs, one committed plan
  • MVP boundary definition separating delivered scope from postponed functionality
    Postponed work stayed visible, not buried
INTERFACE STRATEGY

De-risk the interface decision without reopening the implementation scope.

The inherited UI was functional, but it had limited internal support and did not align well with YPF’s customer-facing products. Replacing it outright would have introduced delivery risk.

I proposed a white-label frontend foundation that let both directions share the same structure and behavior:

  • the inherited theme could still satisfy the committed implementation,
  • a second theme could align more closely with YPF’s existing visual language,
  • and both could be compared without blocking Engineering.

Once the two versions were ready, I presented them in a live demo. YPF selected the alternative direction—the one later deployed in stations.

The important decision was not the visual refresh itself. It was creating a way to improve the product without forcing the organization to choose between design quality and delivery certainty.

Side-by-side comparison of the inherited interface and the proposed YPF-aligned theme
Same system behavior, two interface directions — a reversible decision until YPF validated the stronger one.
  • Proposed kiosk interface — starting the self-service session
    Identify
  • Proposed kiosk interface — selecting fuel and amount
    Select
  • Proposed kiosk interface — confirming and authorizing payment
    Authorize
  • Proposed kiosk interface — fueling in progress and completion
    Fuel and complete
SYSTEM NOTE

AI supported the operation without becoming another customer-facing concept.

The solution used ESPERANTO, a private AI model built with AWS capabilities available at the time, alongside services such as Amazon Rekognition and custom SageMaker components.

From the customer’s perspective, that sophistication stayed mostly invisible. The interface remained organized around the fueling task; infrastructure automation did not become extra UI to understand.

Sanitized architecture diagram of the AI services behind the system
Operational intelligence stayed behind the task model.
OUTCOME

A deployed interface with measurable operational impact and unusual longevity.

The deployed system produced shorter, more transparent waiting in the self-service flow, and greater process transparency for customers and station personnel alike.

The interface direction YPF selected through comparative validation became the production experience, and the core interaction model has remained largely unchanged for years.

The original March 2020 rollout was interrupted by the pandemic. Before moving into a Brand Lead role at Edrans, I transferred the product context and implementation decisions to the designer who continued the next phase. The system was subsequently deployed using the direction I had proposed.

  • Deployed self-service kiosk in production at a YPF station
    Deployed kiosk
  • Deployed self-service kiosk in production at a YPF station
    Station environment
  • Deployed self-service kiosk in production at a YPF station
    Production interface
  • Deployed self-service kiosk in production at a YPF station
    Self-service in use

Get in touch

Want the full, NDA-safe walkthrough? Happy to talk it through.

hello@gastonlasarte.comView my CV
FromBuenos Aires, Argentina
34.6°S 58.4°W
YPF

Product Design