Francisco Pacheco — Senior Product Designer

Francisco Pacheco · Senior Product Designer

Trade data automation
from manual checks to smart compliance

RoleSenior UX Designer PlatformSALOG ScopeTrade Control TeamAIR-04 Agile DurationSince Oct 2025

TL;DR

Background

KN moves cargo across 100+ countries. Every shipment must be checked against trade regulations — sanctions, embargoes, export controls.

Problem

Operators ran compliance checks manually, across 40 data points per shipment. No automation, no audit trail, no consistent enforcement.

Approach

Feature Canvas workshop to align stakeholders. Three research sessions with operators across EMEA, AMER and APAC. Four production releases.

Outcome

7,681 shipments flagged. 2,257 confirmed defence cases. 419 users across 20+ countries.

01 — The Why

A complex problem

Kuehne+Nagel moves cargo across 100+ countries. Every shipment that crosses an international border must be checked against trade regulations — sanctions, embargoes, export controls. Getting it wrong means blocked cargo, fines, or legal liability.

THE RULES

The Complexity

Trade regulations change constantly. KN maintains a living requirements document — currently at version 200 — that defines exactly what operators must check on every shipment.

THE REALITY

The Problem

Operators across Air and Sea run thousands of shipments a day. 99.5% are standard. But that 0.5% — potential defence cases — carries full regulatory exposure if missed or mishandled.

40 data points per shipment Checked manually on every shipment

4 compliance validations GMO · DMC · Sanctions & Embargoes · Export CC

5 layers per validation Identification · Action · Blocking · Approval · Audit

17 prohibited destinations Countries under active sanctions or embargoes — hard block, no exceptions

16 restricted destinations Require special approval before departure — conditional, not blocked

03 — The Iteration

Workshop → Research → Four releases.

The conversation that set the direction.

Before writing a single spec, I facilitated a Feature Canvas workshop with developers, POs, and business analysts — forcing the room to agree on a root cause, a scope, and a solution direction. That board became the shared contract. Everything that shipped traces back to it.

THE PROBLEM

Operators are making compliance decisions from memory.

No system guidance. No structured check. When a shipment looks like it might involve a defence customer or sanctioned goods, it falls on whoever is handling it to know what to do. The root cause the team landed on: human error by design.

PARKING LOT

What we agreed to leave out.

Quoting Internal awareness Language localisation Timeline constraints

Naming what's out of scope is as important as naming what's in. The parking lot kept the workshop focused and gave the team permission to move forward without resolving everything.

SCOPE — 4 SITUATIONS

1.Government Military Organisations

2.Defence Commodities & Customers

3.Sanctions & Embargoes

4.Export Customs Clearance

These 4 mapped exactly to the Business Requirements. The workshop found the scope before the spec existed.

WHO WE DESIGNED FOR

Operators OCC KAMs Business Owners Gateway

The primary users were operators. The broader map — KAMs, Business Owners — was noted but deliberately deferred.

WHAT USERS NEEDED

Know which shipments are affected without having to guess
See what was checked — and what the outcome was
Reduce legal exposure without slowing operations down

03 — The Iteration

Workshop → Research → Four releases.

Research that changed the design.

Three sessions. Nine operators. Sea and Air, across three regions. Each session had one question — and each answer left a mark on what shipped.

Session 1 — Focus Group

Strategy & AI

4 participants · Sea + Air · EMEA · APAC

Evaluate AI flag vs. block authority model

Session 2 — Focus Group

Visibility

4 participants · Sea + Air · EMEA

Where to surface AI alerts in the workflow

Session 3 — Individual Interview

US Operations

1 participant · Air · AMER

Validate TC workflow with high-frequency defence operator

3 Sessions Strategy & Visibility

9 Operators OCC · CCL

3 Regions EMEA · AMER · APAC

2 Domains Sea · Air

Observation 01

Operators prefer flag-based control over hard blocks.

Session 1 · Sea/LCL · EMEA & APAC

Observation 02

AI decision data is invisible to most of the team — visibility is as important as the alert itself.

Session 2 · Sea · EMEA

Observation 03

High-frequency operators already self-verify. The tool needs to support that workflow, not replace it.

Session 3 · Air · AMER

03 — The Iteration

Workshop → Research → Four releases.

Four releases.

Visibility

Trade Control Tab

Blocked Actions

AI & Analysis

Sanctions

R4.30

Nov 2025

● Shipped

Party labels

Defence Customer & GMO labels in Parties tab + shipment header

R4.32

Mar 2026

● Shipped

Header chips

Defence / Potential Defence / Blocked chip variants on shipment header

Commodities + Manual Release

Defence Commodities field + entitlement-gated Manual Release

Process

AWB / MAWB / Carriers

Automatic blocks on AWB set, MAWB closure, and forbidden carriers

R4.33

May 2026

● Shipped

Events table

Trade Control events log visible on tab

Process

Mandatory check

Mandatory commodity check for Potential Defence + BPA auto-task

R4.34

Ready for deployment

● Ready

Header chip update

New blocked state for Sanctions & Embargoes flag on shipment header

Defence Shipment Info

Principal party + Cargo Description + KN Flow Number on tab

Process

Sea doc block

Block document production for Blocked Defence — Sea

Sanctions Red Flag

Raise / lower red flag + blocked actions + visible in header

Future

Roadmap

In design

Process

Schedule B validation

US compliance flag for European transit

Process

Routing restrictions

Continuation of routing logic for Defence shipments

Document Analysis

AI reads defence docs, auto-classifies + surfaces findings on tab

AI full rollout

Document Analysis extended to Sea & Air

05 — Impact

Shipped. Measured. Real.

Four releases across two product areas. Numbers pulled from production — not projections.

Memory-based compliance checks 40 data points verified manually No structured audit trail False blocks flooded the Defence team

The numbers so far.

7,681 Shipments flagged

2,257 Confirmed defence (29%)

334 Auto-blocked GMOs

419 Active users

20+ Countries active

R4.30 Defence Party Recognition Nov 2025 ● Shipped

The system shows the label immediately in the Parties tab and shipment header. Operators no longer make judgement calls.

Shipment Header
1 — Dark colour labelsNo lookup required.
Parties Tab
2 — Party type inlineDCU / GMO labels shown per party card. Operators see it at a glance before deciding.

R4.32 & R4.33 Defence Shipment Recognition & Events Mar – Apr 2026 ● Shipped

The Trade Control tab brings auto-detected Defence shipments, commodity confirmation, and the Manual Release flow into a single view.

Shipment Header
1 — Classification chipPot. Defence / Defence Shipment / Blocked Defence — status visible at a glance before opening the tab.
Trade Control Tab
2 — Consolidated viewAll trade compliance signals in one place — previously spread across multiple screens.
3 — Commodity listDetected defence-related commodities surfaced with classification and controlled-goods indicator.
Event History
4 — Events tableChronological log of all trade-control decisions — visible to ops and compliance teams.

Manual Release slide-in
1 — Guided formOperator override in a structured slide-in — not a free-text field.
2 — Reason captureEvery release decision is auditable and tied to a specific reason.

R4.34 Defence Shipment Information · Sanctions & Embargoes Q2 2026 ◐ Ready for deployment

Design and development complete. Waiting for deployment window. Two additions to the Trade Control tab — contextual information for operators, and a manual red flag mechanism for anything that doesn't look right.

Shipment Header
1 — Red flag visibleActive flag state surfaced at the top of the shipment — unmissable for anyone who opens the file.
Defence Shipment Information
2 — Principal party cardThe key commercial party surfaced directly on the tab — no switching screens.
3 — Cargo descriptionVisible in context, next to the commodity classification. Operators can cross-reference without leaving the tab.
Sanctions & Embargoes
4 — Forwarding actions blockedOnce flagged, the shipment cannot move forward until cleared by an authorised operator.

Raise & lower slide-in
1 — Reason captureEvery decision is auditable — operator must state why the flag was raised or lowered.
2 — Local accountabilityThe decision to lift the flag stays with the operator who knows the shipment — not a central review team.

Future Schedule B, full AI rollout, and what comes next — covered in the next section. See Beyond this case study →

06 — Beyond this case study

Designing with AI,
not just for it.

Not a tool. A workflow.

I've been compressing the product design cycle by integrating AI across eight areas — from spec generation to design system maintenance. Less repetition, more decision-making.

Product Design

Active

Second Brain

AI-managed knowledge system that never goes stale

Obsidian vault + AI agents that process meeting notes, extract decisions, and keep project context always current. Every session starts with full context.

ObsidianClaudeAtlassian

Product Design

Active

Spec Generation

From meeting notes to dev-ready specs in one pipeline

Agent pipeline processes tickets, transcripts, and Figma annotations into structured HLS specs. I edit and decide — not write from scratch.

ClaudeObsidianAtlassian

UX Design

Active

Design Sparring

AI as a thinking partner, not a yes-machine

Discuss patterns in plain language. Ask for the case against. Force a recommendation. Decisions and discarded alternatives end up in the decisions log.

ClaudeObsidian

UI Design

Active

Rapid Prototyping

Interactive HTML prototypes in hours, not days

Using Claude to generate interactive prototypes for stakeholder validation — including this case study — before committing to full Figma builds.

Claude

Design System

In Progress

DS as Constraint

Design system as guardrail, not just a library

Encoding tokens, hierarchy rules, and component patterns into a DESIGN.md file the agents read before generating anything. DS compliance from prompt zero.

ClaudeStorybook

UI Design

In Progress

AI → Figma Canvas

Bringing AI-generated UI into the design file

Generate UI fragments, write them to Figma via the official Figma MCP (write-to-canvas, beta), refine against design-system tokens, ship to dev.

ClaudeFigma

Design System

On the Radar

DS from Production Code

Auto-generate the Figma library from what actually ships

Scan production codebase for tokens, styles, and component APIs; auto-generate a Figma library that mirrors the live product. No more drift.

ClaudeFigmaGitLab

Product Design

On the Radar

Analytics Feedback Loops

Continuous agent-driven insight from product data

Agents ingest product analytics, identify friction patterns, and surface actionable feedback into the design vault automatically — without waiting for a research sprint.

ClaudeObsidian

In-use Exploring Waiting

This taxonomy of capabilities was articulated by Patrick Kluge Milani at redesignthis.org (2026). The grid above maps which categories my method addresses today.

How to implement

Days 1–30

Audit and listen.

Map the design-ops baseline across three pillars: where specs come from, how research lands, where the design system lives, where the friction loops are. No tool changes yet.

Days 31–60

Codify constraints.

Take the existing design system and make it a guardrail, not just a library — encoded in a form an agent can read before it generates anything.

Days 61–90

Pilot one agent.

Identify the highest-friction intake-to-spec loop. Build a scoped agent for it. Measure baseline first, measure after. Honest result, not a fabricated number.

Protected case study

A complex problem

The Complexity

The Problem

How the team is structured

The conversation that set the direction.