Designing the Future of Enterprise Finance Setup and Configuration
How I led the design of a unified ERP, EPM, and EDM foundation service by grounding an AI-first vision in UX research, generative experimentation, and design system innovation
The final exam for my Nuclear Electronics Technician program was a true test of not only my knowledge of electrical systems, but even more of my troubleshooting abilities. I was presented with a radar system that was malfunctioning and given two hours to identify the issue and fix it. I had never before even seen a radar system.
Troubleshooting something you’ve never encountered before is a tall order, but the process remains the same as always. I quickly surveyed the radar screen and controls to understand how it should work, then began diving into its internals to understand what makes it work. I began testing signals and components methodically, identifying patterns and relationships, so I had a clear picture of precisely how the system worked and the sub-processes that operated to achieve that.
Once I had that picture, I was able to step back and swiftly isolate the issue and formulate a plan to resolve it. I reviewed what I had available to accomplish that and chose the best course or action.
In less than two hours I had fixed the radar system and could identify ships off the distant coast of Florida, as well as cars in the parking lot outside the training center. I passed.
Being able to quickly grasp a system, its elements, and the patterns that both make it work and indicate a problem, has remained an essential skill that I keep in my designer toolbox.
Project Asimov
The product owner’s directive was bold: eliminate consultants from corporate finance app implementations and replace them with a natural language mechanism that could rapidly create enterprise structures and dimension members. But that wasn’t the only ambition. Asimov also had to hide the seams between ERP, EPM, and EDM so customers felt they were using a single product rather than three siloed applications. And finally, from the very first step, it needed to centralize key data such as dimensions, naming, and hierarchies, solving a problem that had plagued enterprises for decades.
Historically, ERP and EPM teams had worked in parallel, building their own structures and naming conventions that constantly drifted out of sync. Customers spent enormous effort reconciling mismatched data, often relying on consultants to untangle the mess. Asimov promised something different: a unified system that started with shared data and kept it consistent throughout its lifecycle.
I was brought in as lead designer to clarify this radical vision and turn it into something practical. My role was to define the experience, ground it in research, and begin shaping both the product flows and the design system patterns that would make it real. While I did not stay with the project long enough to see the MVP released, my work established the foundation and the vision that will guide the team toward launch.
The Challenge
Replacing consultants while unifying systems and data
When I joined, engineering had early experiments with conversational agents, and product leadership was enthusiastic about natural language interfaces. But it became clear that the core challenges weren’t about chat or UI polish.
Administrators told us they wanted an easy way to setup and configure financial and legal data that eliminated the headache of data mapping management. Finance leaders worried about compliance and auditability; the idea of a black-box AI making structural decisions made them uneasy. Change Champions pointed out that they were already working in spreadsheets and documents, so forcing them to re-enter information into a new flow would only create frustration and errors. And above it all, executives insisted that ERP, EPM, and EDM should not feel like three different systems stitched together, but a single, coherent product driven by consistent, centralized data.
The challenge wasn’t to create an AI assistant. It was to design an intelligent foundation service that delivered consultant-level confidence, unified siloed systems, and created a single source of truth for enterprise data that demanded less expertise in a myriad of complex configuration UIs.
The Early Vision for the MVP
Anchoring on enterprise structures and dimensions
To make progress, we narrowed the focus to one critical outcome: the creation of enterprise structures and initial dimension members in greenfield implementations. This was the choke point where consultant costs ballooned and where errors carried downstream consequences for every app and process.
The first release was framed around a narrative: customers would begin not by filling out endless forms, but by providing documents they already had—narrative corporate structure and documents. These would be ingested and interpreted by the system. From there, the admin would take the role of orchestrator, coordinating inputs from Change Champions and finance teams, while the AI guided the process, surfaced issues, and produced a narrative/structured “blueprint” ready for validation and deployment.
It was a pragmatic scope. Customers would not yet see the product—launch was still years away—but the story was clear. The MVP wasn’t meant to solve every problem at once. It was meant to prove that consultant-free, cross-system setup could be both credible and trusted.
High-level model of the system, illustrating how the administrator acts as the primary liaison between the AI and the financial team Change Champions, orchestrating the inputs and outputs to achieve a final, acceptable setup and configuration blueprint.
Experimenting with Generative AI
Testing the hypothesis with real documentation
To validate feasibility, I conducted extensive generative AI experiments:
Uploaded and curated real-world corporate documents and files to simulate inputs.
Measured outputs to test whether AI could generate valid, configuration-ready structures across ERP, EPM, and EDM.
Mapped weak points where AI produced inconsistencies, then began designing flows that layered in human review, trust signals, and transparency.
Benchmarked against traditional four-prototype implementation phases, showing potential to eliminate one or two iterations.
These experiments directly addressed a growing obstacle: project managers and executives couldn’t pin down the nature, number, order, or content of documentation needed to produce a solid blueprint for ERP, EPM, and EDM configuration. Every week brought new uncertainty. I cut through the noise with real use-cases, showing exactly what each document could deliver and how generative AI, paired with the right UX patterns and governance, could accelerate foundation creation.
I also proved we could lower user effort even further. Instead of asking customers to supply corporate data up front, our agent could auto-collect it from public sources like the SEC. This flipped the experience: users began with a blueprint already prepared for review—seemingly “by magic”—rather than facing an initial data-entry hurdle. Early tests validated the approach, and we reshaped our workflow around it.
Exploration of genAI prompts and inputs/outputs
Advancing Redwood
Designing new patterns for generative AI
The experiments also revealed a gap: the design system wasn’t yet prepared for leveraging generative AI or cross-application governance in this way.
I documented requirements and began designing patterns that would allow Redwood to support this new class of experience:
Transparency – showing AI’s reasoning, not just its results.
Trust-building – confirmations, checkpoints, and audit trails.
Control – allowing users to refine or override AI outputs without breaking flows.
Clarity in language – structuring natural language interactions to reduce ambiguity in financial contexts.
Cross-product consistency – unifying ERP, EPM, and EDM through shared components and flows.
These weren’t just design tweaks. They represented an evolution of the design system itself, expanding its reach from traditional enterprise UI to AI-first, document-driven experiences for some of our users’ most challenging obstacles.
Expanding the Vision
From setup accelerator to unified enterprise foundation
With the foundation experience defined, I worked with product leadership to describe a longer-term vision. In this future, silos between ERP, EPM, and EDM disappear. Customers create structures and dimensions once, and they remain consistent everywhere. Data flows seamlessly across applications, backed by transparency and governance. And most importantly, users experience this not as a collection of tools, but as a single enterprise foundation service.
This north star remains a vision as of this writing. Customers have not yet used the product, and the MVP is still far from delivery. But it is a vision with intent, clarity, and executive alignment.
Photo by Johannes Plenio on Unsplash
The Outcome (So Far)
A vision that redefines ERP delivery
While the Asimov MVP has not yet reached customers, my work created:
A vision and scope for greenfield setup that eliminates consultant dependency.
Evidence from generative experiments that accelerated cycles and validated feasibility.
Documented requirements and patterns to evolve the design system for AI and cross-application governance.
A roadmap for unifying ERP, EPM, and EDM into a single product experience.
Executive confidence that UX leadership could translate ambitious directives into a credible product strategy.
My requirements for AI UX
What I Learned
Project Asimov reinforced that UX leadership is not only about designing screens, it’s about defining vision, validating feasibility, and setting up teams for long-term success.
Three lessons stood out:
Generative AI is a design research accelerant – experiments with real documents revealed both potential and limitations.
Design systems must evolve with AI – the design system required new patterns for trust, clarity, and governance to support enterprise AI.
Vision matters even without delivery – defining the north star for consultant-free, unified ERP/EPM/EDM laid the groundwork for business and design success.
Project Asimov is still on its journey, but the foundation has been set. The directives are clear, the vision is defined, and the design system is evolving. When it reaches customers, they won’t just see another ERP tool. Instead, they’ll experience a unified enterprise foundation service that redefines how implementations are done.