ANNA
EUSTAQUIO
AI-Powered Reconciliation Engine
Global Compliance Overview
Fintech
ERP
AI UX
Year
2025-2026
Timeline
8 months
Industry
Finance, P2P
Role
UX Designer
Team
1 PM, 4 Engineers
Industry
Finance, Localization
Role
Lead Designer
Team
1 PM
Year
2025
Timeline
48h
Industry
Design Ops/ AI for designers
Role
AI Design Facilitator
Team
10 designers
Year
2025
Timeline
48h
Translating dense business logic into high-performance interfaces.
Specialized in fintech, compliance, and AI-assisted workflows.
About me
I’m a UX Designer with a background in linguistics and teaching, which shaped how I navigate structure, communication, and cognitive load. Today, I apply those learnings to the world of complex enterprise and software.
I specialize in the high density environments of UXERP, transactions, and automation tools. I approach design as a systems problem: my goal is to translate dense business logic into clear interfaces that prioritize accuracy and speed.
Whether I’m designing AI-assisted tools or simplifying high-stakes approval flows, I focus on reducing ambiguity and enabling human oversight.
I specialize in the high density environments of UXERP, transactions, and automation tools. I approach design as a systems problem: my goal is to translate dense business logic into clear interfaces that prioritize accuracy and speed.
Whether I’m designing AI-assisted tools or simplifying high-stakes approval flows, I focus on reducing ambiguity and enabling human oversight.
Selected work
AI-Powered Reconciliation Engine
NetSuite
Fintech
ERP
AI UX
Global Compliance Overview
NetSuite
ERP
AI UX
Sprint
Figma AI Team Training & Consultancy
NetSuite
AI UX
DesignOps
Prompt Engineering
E-Sign Flow for Insurance Conversion
UX Challenge · Fintech
Fintech
Conversion UX
Compliance
Contact
If you want to learn more about my approach to logic and UX, let’s have a chat.
Email
aeustaquiobaixauli@gmail.com
LinkedIn
/in/annaeebb
Overview
In most finance departments, "reconciling" is just a fancy word for a tedious, manual game of "spot the difference." Even with ERP systems like NetSuite, teams spend hours every day cross-referencing purchase orders, bills, and bank statements.
When the numbers don't match exactly, automation usually fails. This forces a human to step in and manually hunt through different platforms to figure out why. It’s a slow, exhausting process that leads to burnout and easy-to-miss mistakes.
When the numbers don't match exactly, automation usually fails. This forces a human to step in and manually hunt through different platforms to figure out why. It’s a slow, exhausting process that leads to burnout and easy-to-miss mistakes.
UX Challenge
Designing a financial reconciliation engine means balancing messy data with the crucial need for accuracy. Key factors to consider when designing this solution include:
Designing a financial reconciliation engine means balancing messy data with the crucial need for accuracy. Key factors to consider when designing this solution include:
Delivered under an emergency timeline of less than 48 hours, this dashboard was presented live on stage at Oracle NetSuite's annual executive keynote in Las Vegas to an audience of 6,000+ customers, partners, and industry stakeholders.
The design had to communicate multi-country tax compliance status at a glance, without any prior prototype existing. Working directly from compliance requirements and leadership briefs, I produced a stage-ready, high-fidelity dashboard that visualised regulatory status across jurisdictions in a single view. Presented live to the entire design organization, this session showed that a structured prompt and a disciplined iteration strategy can get a team to design alignment significantly faster than traditional methods.
The session changed how the team approaches AI, moving from experimentation to a repeatable, documented methodology. The TC-EBC framework I introduced became the internal standard for Figma AI workflows, turning Figma Make from a random visual generator into a predictable, professional validation tool.
After the presentation, multiple product teams reached out independently to bring the methodology into their own design processes. This project showed that structured prompt engineering is not just a productivity trick. It is a design systems problem, and it benefits from the same rigour.
The design had to communicate multi-country tax compliance status at a glance, without any prior prototype existing. Working directly from compliance requirements and leadership briefs, I produced a stage-ready, high-fidelity dashboard that visualised regulatory status across jurisdictions in a single view. Presented live to the entire design organization, this session showed that a structured prompt and a disciplined iteration strategy can get a team to design alignment significantly faster than traditional methods.
The session changed how the team approaches AI, moving from experimentation to a repeatable, documented methodology. The TC-EBC framework I introduced became the internal standard for Figma AI workflows, turning Figma Make from a random visual generator into a predictable, professional validation tool.
After the presentation, multiple product teams reached out independently to bring the methodology into their own design processes. This project showed that structured prompt engineering is not just a productivity trick. It is a design systems problem, and it benefits from the same rigour.
AI Trust Gap
The typical user is an auditor who is naturally skeptical. Because AI works with probabilities and is not always 100% absolute, the UI must clearly communicate its reasoning and confidence levels. If the system matches data without an explanation, users will lose trust and revert to manual checking.
Cognitive Load
Data is often spread across different platforms and formats. Forcing users to jump between them to verify a transaction creates a significant mental burden. The core challenge is consolidating all of them into a single, accessible format.
Cost of Error
In a finance environment, "close enough" is not an option. Data must be completely accurate, as a minor UI error could result in significant compliance issues or financial loss.
The Process
Ideation and alignment
I kicked off the project by leading cross-functional workshops with engineers and PMs. We used Crazy 8s to rapidly explore diverse ways of visualizing complex financial data and to align the team’s technical constraints with user needs.
Chatbot configuration
Data mapping & edge cases
I performed a deep dive into the reconciliation logic to understand how different sources, like purchase orders, vendor bills or bank statements, interact. I focused on identifying and solving all edge cases, ensuring the system could handle all possible discrepancies.
Structured Prompt Engineering
AI interaction and patterns
I researched and applied AI UX patterns specifically focusing on trust and transparency. I designed the interface to move away from "black box" automation, focusing on how to communicate probability and logic so users felt confident in the AI's suggestions.
Flow Validation with Figma Make
High-fidelity designs
Once the logic was solidified, I moved into high-fidelity UI design. I focused on reducing cognitive load by creating a dense but scannable dashboard that unified fragmented data into a single, accessible format.
Visual Iteration
Fast validation
To finalize the direction, I used Figma Make to build an interactive prototype. This allowed us to validate different ideas and workflows in real-time, using AI as an accelerator to refine the user experience and reach a final team consensus.
Data unification flowchart
Impact
This project is currently in active development. The reconciliation engine was designed to eliminate the manual cross-referencing of purchase orders, vendor bills, and bank statements.
The interface was built to handle ambiguous matches that would previously require a human to hunt across multiple platforms. By surfacing confidence scores and reasoning paths inline, the design reduces context-switching and keeps the auditor's decision-making centralised in a single view.
The framework established during this project, AI suggestions with traceable logic and mandatory human sign-off on low-confidence matches, is being carried forward as the UX standard for AI-assisted finance workflows within the product.
The interface was built to handle ambiguous matches that would previously require a human to hunt across multiple platforms. By surfacing confidence scores and reasoning paths inline, the design reduces context-switching and keeps the auditor's decision-making centralised in a single view.
The framework established during this project, AI suggestions with traceable logic and mandatory human sign-off on low-confidence matches, is being carried forward as the UX standard for AI-assisted finance workflows within the product.
48-Hour Sprint
The timeline was the biggest hurdle. I delivered the first draft in 4 hours and the final, stage-ready version in 2 days. This required making high-level decisions
High-importance project
Because this was a VP-level centerpiece for a global convention, it had to be perfect. I coordinated with other designers and stakeholders, managing a constant feedback loop to ensure the solution was technically accurate and followed our design system.
Mastering Legal Complexity
I had to learn international tax mandates on the fly. The challenge was finding a single design pattern that could represent dozens of different localized laws, turning fragmented data into a clear "Global Health" status.
Key Takeaways
My approach for this project focused on bridging the trust gap through transparency, ensuring that while the AI handles the heavy lifting, the human remains the ultimate authority.
Confidence over Automation
Automate the obvious, flag the ambiguous. By leading with confidence scores, the UI lets auditors ignore the "perfect" matches and focus energy solely on the high-risk exceptions where things actually break.
Traceability by Default
Never show a result without the "why." Every match includes a reasoning path that links specific data points between POs and bills, giving auditors the evidence they need to trust the AI's logic
Human-in-the-Loop Authority
In high-stakes finance, speed isn't the only metric. I built in manual verification steps for low-confidence matches to ensure the user stays the final authority and avoids costly errors.
Unified Data Mapping
Cut the cognitive load by mapping fragmented data into a single flow. Bringing disparate formats into one view means auditors don't have to jump between platforms to verify a transaction.
Overview
This project started with a last-minute request from leadership: we needed a visual design for an executive keynote in Las Vegas, with more than 6000 atendees. The goal was to showcase a new Global Compliance Dashboard that hadn’t been prototyped yet.
For multinational companies, staying compliant is a headache. Tax rules are fragmented across borders and each country has specific rules and timelines. I led an emergency design sprint to turn this legal complexity into a clear, "stage-ready" dashboard that showed leadership exactly where they stood globally.
For multinational companies, staying compliant is a headache. Tax rules are fragmented across borders and each country has specific rules and timelines. I led an emergency design sprint to turn this legal complexity into a clear, "stage-ready" dashboard that showed leadership exactly where they stood globally.
Overview
Most designers treat AI as a random generator. They input a "guess" and hope for a useful result. This creates a cycle of trial and error that wastes time and makes it impossible to use AI for actual product alignment. I saw the need to move the team from "exploring" to "executing" by creating a reproducible, scientific way to talk to the AI.
I presented to our company-wide design organization to transition the team from exploring to executing with a standardized workflow for Figma AI. Rather than treating AI as a random visual generator, I showed how it can act as a high-velocity validation tool when approached scientifically. To ensure every prompt was technically complete and aligned with our product goals, I introduced the TC–EBC framework (Task, Context, Elements, Behavior, and Constraints) moving us away from a cycle of trial and error toward a reproducible, professional standard for prompt engineering and workflow optimization.
I presented to our company-wide design organization to transition the team from exploring to executing with a standardized workflow for Figma AI. Rather than treating AI as a random visual generator, I showed how it can act as a high-velocity validation tool when approached scientifically. To ensure every prompt was technically complete and aligned with our product goals, I introduced the TC–EBC framework (Task, Context, Elements, Behavior, and Constraints) moving us away from a cycle of trial and error toward a reproducible, professional standard for prompt engineering and workflow optimization.