Client logo

Building AI tools to visualize risk and streamline corporate processes

Role

Sr. Product Designer

Timeline

12 weeks

Platforms

Web

Final product view

Overview

Systematizing Intelligence

AI in Due Diligence processes for PwC: I led the conceptualization of a modular component ecosystem to automate global audit processes. By standardizing complex patterns and optimizing usability, we accelerated development and ensured organic adoption across critical workflows.

Key Results

+31%

Of early adoption by flattening the learning curve and building trust in the system's predictions.

2.3x

Streamlined due diligence processes, significantly reducing execution time for auditing teams.

99%

Ensuring technical and visual consistency of AI-driven components across global brand guidelines.

+6

Design, testing, and integration of micro-frontends into shared libraries to power scalable internal workflows.

The Situation

Optimizing demanding workflows

A Due Diligence process involves deeply auditing a company's risks prior to a multi-million dollar transaction. Operating in a zero-error environment, auditors relied on manual scrutiny of massive data sets. The goal was to integrate AI to optimize the workflow, but the strategic gap lay in defining how to do it without introducing friction into such a critical and sensitive operation.
Files icon

1. Collection

Requesting and uploading hundreds of documents for each audited company.

File classification

2. Classification

Manually naming and categorizing the massive volume of files on the platform.

File analysis

3. Analysis

Intensive, line-by-line reading to extract information and answer questionnaires.

File diagnosis

4. Diagnosis

Identifying legal anomalies and transferring those findings into a risk matrix.

Audit report

5. Reporting

Consolidating detected issues to generate the final executive document.

Discovery

The Human Factor

I interviewed audit teams to identify where AI added the most value. We discovered that technology wasn't the challenge, but rather three critical adoption barriers:

The "Black Box" Fear

Approx. 40% distrusted AI’s judgment or feared being replaced. To validate a conclusion, users required full visibility into the machine’s reasoning.

Traceability: The Foundation of Trust

Adoption relied on evidence. Every prediction or summary had to include exact citations and direct links to the original document; without a source, there is no validation.

Context Switching Fatigue

Auditors rejected external chats or new tabs. Assistance had to occur natively within their workspace to avoid fracturing their reading flow.

Chart of findings on AI adoption barriers

Discovery

Analyzing the tools

We discovered that an isolated conversational module doesn't eliminate the operational burden. Auditors didn't need to answer questions somewhere else, but rather an AI that automated processes directly within their workflows.

Comparative visual analysis of AI tools

Define

How Might We...

...integrate AI into processes so it acts as a reliable copilot without disrupting current workflows? To solve this, I facilitated a workshop with Business, Product, and Tech leaders, where we defined the three strategic pillars that would shape this new ecosystem:

AI blocks visualization

Transparency

The system must guarantee traceability and full visibility for every piece of information.

Trust

The design must protect expert autonomy; the AI proposes, the auditor decides.

Scalability

A modular extension of current tools, built to scale without friction.

Ideation

Assembling the lego

Working alongside a group of auditors, we defined critical activities that, by nature, always required human review. This allowed us to isolate time-intensive tasks where AI could truly streamline the workload.

Diagrama modular separando tareas de revisión humana de tareas automatizadas por IA

Ideation

The AI Visual Standard

To accelerate adoption, I designed an exclusive visual identity for AI components with a dual purpose:

For the Auditor

Visually differentiating AI suggestions from native data to shorten the learning curve and prevent confusion during analysis.

For PwC (Scalability)

Establishing a design standard so other teams can integrate AI into their own projects, ensuring a consistent experience globally.

Wireframe conceptual sketch of the AI logo
Volumetrics and geometry of the AI logo
Final color version of the AI visual standard

UI Design

Component System

I architected a modular component ecosystem that enables auditors to interact with AI at any point in their workflow, while ensuring full traceability at all times.

AI Statistics
Questionnaire
Voice Input
AI-Powered Search
AI Classification Tags
Detailed Findings
Chat Assistant
Analysis Chart

Delivery

Functional Implementation

We converted the modules into configurable micro-frontends. This approach enabled engineering to implement AI natively within the workflow, eliminating friction and allowing for scalable adoption.

Key Takeaways

Design as Governance

In global-scale organizations, design isn't just interface; it's policy. Creating a visual standard for AI unified technical and aesthetic criteria, facilitating seamless adoption across multiple teams within the firm.

Transparency is the new UX

In high-risk environments like Due Diligence, aesthetics are secondary to traceability. I learned that to overcome technological resistance, design must prioritize evidence and data provenance over total automation.

Systems over Screens

Solving technical fragmentation required shifting from isolated views to designing a modular ecosystem. The product's success lay not in "what the AI did," but in "how it was natively injected" into the pre-existing workflow.

Thanks for sticking around!

If you enjoyed the process or want to dive deeper into any part, let’s chat.

Get in touch