Designing Decision Governance in Regulated AI Systems

Wells Fargo · Lead Design Thinker, AI Enterprise Solutions · 2019

---

Context

Wells Fargo’s AI Enterprise Solutions group was tasked with advancing machine learning across fraud detection, claims processing, and operational systems. Despite significant technical capability, adoption lagged.

The barrier was not technology. It was trust.

Business units operated in a deeply risk-averse environment, where experimentation was often interpreted as exposure. AI initiatives stalled due to regulatory uncertainty, unclear ownership, and lack of alignment between legal, data science, and product teams.

---

The Problem

AI efforts failed not because models were ineffective, but because no system existed to make their outputs accountable.

Teams encountered:

• inability to frame AI initiatives in compliance-safe terms
• misalignment between technical capability and regulatory expectations
• lack of shared language across legal, product, and data teams
• absence of structured pathways from idea → experiment → approval

The organization had AI ambition, but no reliable way to operationalize it.

---

Intervention

I led design strategy to create decision governance frameworks that made AI experimentation structured, observable, and compliant.

The focus was not on building models, but on designing the system around them.

These frameworks introduced friction intentionally, requiring teams to define risk, ownership, and decision boundaries before experimentation could proceed. This shifted AI work from informal exploration to accountable system design.

Key changes included:

• Decision framing templates
  Standardized how teams defined assumptions, risks, data requirements, and expected outcomes, translating abstract AI ideas into testable hypotheses.

• Compliance-aligned experiment canvases
  Enabled teams to validate machine learning use cases within regulatory boundaries, reducing friction with legal and governance stakeholders.

• Cross-functional labs
  Created shared environments where risk officers, data scientists, and product leaders could collaborate on controlled experiments.

• Human-in-the-loop checkpoints
  Embedded validation, escalation, and override mechanisms into early-stage AI workflows to ensure accountability.

• Regulatory trust layer
  Introduced interpretability thresholds, audit scaffolding, and decision traceability into the design of AI outputs.

---

Constraints

The work operated within:

• strict regulatory and compliance requirements
• high organizational sensitivity to operational and reputational risk
• fragmented ownership across business units and technical teams

Any approach to AI had to be credible to legal and risk stakeholders while remaining actionable for product and engineering teams.

This required aligning stakeholders who were structurally incentivized to avoid risk rather than manage it, creating tension between innovation velocity and institutional accountability.

---

Outcome

AI shifted from a theoretical capability to a structured, testable discipline.

Observed outcomes included:

• 4 AI pilots launched within 3 months across fraud and compliance workflows
• re-evaluation and approval of previously stalled models, enabling deployment pathways
• increased stakeholder confidence in AI initiatives (qualitative: “Feels safe to try”)
• cultural shift from risk avoidance to structured experimentation

Critically, teams that previously could not initiate AI work due to regulatory ambiguity were now able to frame, test, and advance proposals within defined governance boundaries.

---

Result

Wells Fargo moved from treating AI as a risk to managing it as a governed system.

The work reframed innovation as a controlled process rather than an uncontrolled threat.

The system shifted from:

“Is this allowed?”

To:

“How do we test this safely, and what does success look like?”

This established a foundation for AI adoption rooted in trust, accountability, and decision clarity.