Understanding the People, Processes, and Pain Points Behind FX Repatriation
| Property | Value |
|---|---|
| 📅 Project | OmniRisk · Financial Workflow Dashboard |
| 🔖 Phase | Discover |
| 🛠️ Method | Semi-structured interviews · Market & secondary research · Competitive analysis |
| 📅 Timeline | Weeks 1–3 of the 10-week project |
The research plan was designed to answer one central question: What conditions would have to be met for sophisticated, skeptical operators in institutional finance to trust and adopt an AI-augmented workflow tool — and where are those conditions currently unmet by existing platforms?
The plan combined four research streams — market studies, secondary research, competitive analysis, and primary user interviews — to produce decision-grade evidence for design, not statistical representativeness across the industry.
One — understand the FX repatriation workflow as it actually runs. Not how it's documented in process manuals or pitched in vendor marketing. How the people who own different parts of it describe living inside it.
Two — surface where the workflow lives in the seams between existing tools. The research goal was to identify, specifically, where the disconnected tools force the analyst to become the integration layer.
Three — identify the conditions that determine whether AI features get adopted or rejected. Institutional skepticism toward AI is not an OmniRisk-specific challenge; it is a structural feature of the market. The research had to surface the adoption conditions — the design features that make AI usable rather than ignored.
Four — capture regulatory and compliance constraints as design baselines. Audit trails, override governance, and reporting requirements (MiFID II, Dodd-Frank) are not features; they are conditions the design must meet to be usable in production environments.
| Theme | Question |
|---|---|
| Workflow | What does the FX repatriation workflow actually look like across the people who own different parts of it? |
| Tooling gaps | Where do existing tools (Bloomberg, Aladdin, Enfusion, FCM portals) fail the analyst running the workflow? |
| AI attitudes | What are the specific conditions under which operators would trust AI recommendations enough to act on them? |
| Override | At what point in the workflow does the operator need the ability to intervene, and how should that intervention be governed? |
| Audit | How is audit trail documentation currently produced, and where does it break down? |
| Cross-office friction | Where does the middle office–front office handoff create timing or visibility gaps? |
Primary research: Semi-structured one-on-one interviews. 60 minutes per session. Conducted remotely via video call with screen sharing where participants wanted to demonstrate their current tooling.
Supporting research: Three secondary streams running in parallel with the interviews: