Turning a Request for “Approval Workflows” into a Governance Engine

Three Tier-1 accounts, each a household name in its industry, demanded centralized financial governance, which reached us as a single, ambiguous mandate: "Uptempo needs approval workflows."

The Problem: We had no formal specifications, user stories, or defined scope. All we had was a static conceptual model from one client and a feature name. Three demanding clients needed it, and a hundred more would inherit whatever we built.

Organizational Alignment: I started by facilitating initial discovery workshops to align Product Managers, establish a unified definition of enterprise governance, and evaluate technical feasibility.

Rather than waiting on nonexistent requirements, I built a baseline prototype for clients to react to. While the initial logic was rough and full of gaps, it successfully communicated the core system mechanics, and our Tier-1 clients enthusiastically validated the foundational approach.

With client availability for follow-up interviews weeks away and sparse Tier-1 data, I led an AI-accelerated Design Thinking discovery sprint to close the gap and keep momentum. By synthesizing industry and internal domain expertise across Product, Customer Success, and Services, I mapped complex edge cases, approver hierarchies, and threshold rules.

Reviewing our discovery models with my Financial Management counterpart revealed a stark contrast: mid-sized clients desperately needed governance, but without the Tier-1 overhead.

This accelerated synthesis translated scattered data into actionable system logic:

With the actors, journeys, and segment requirements locked, we transitioned from discovery into drafting the foundational system architecture.

PMs had to accept changes to their domains, designers had to converge on a unified model, and engineering needed to trust the architecture. I drove this alignment through three principles:

Hardcoding approval logic into individual product modules meant we would have to rebuild it every time a new governance need arose. To prevent this technical debt, I architected a decoupled, centralized engine where administrators could configure custom workflows globally.

We needed to build an MVP without trapping ourselves in a corner. To ensure our early decisions didn't quietly limit the product's future, we designed a 6-to-9-month development roadmap.

Campaign managers submit approval requests directly within their planning activities, attaching contextual evidence rather than isolated numbers. These automatically route to a role-based inbox, giving reviewers instant, end-to-end visibility for fast sign-off.

When managing hundreds of requests a month, governance shifts from individual approvals to proving the system held up over time. We introduced filters, delegation rules, and an immutable history to ensure bulletproof auditability.

I designed a no-code configuration flow where each process links a financial threshold to specific approval steps, allowing admins to configure both the trigger amount and the form fields each step requires.

Enterprise buyers validated this initial design, and early feedback was overwhelmingly positive—until it hit real-world operational chaos.

Approvals scaled into the hundreds, breaking our process-centered architecture. A single organizational change could force admins to manually update dozens of processes.

To solve the scaling issue, I decoupled the workflow engine into a modular, object-based architecture. We broke the system down into three reusable building blocks:

Reusable Templates: We replaced static, hardcoded workflows with flexible templates mapped directly to the client's Uptempo financial hierarchy (Budgets, Cost Centers, and Marketing Activities).

Independent Threshold Entities: We reengineered threshold rules into standalone governance objects, allowing admins to dynamically map multiple approval flows to them.

Shipping an enterprise MVP in three months required aggressive scope discipline. To validate the platform with pilot clients without delaying launch, we consciously deferred non-critical capabilities—using temporary third-party integrations (via Workato) for complex threshold rules until native functionality was built in Phase 3.

A minimal application form capturing only the core business case and deadline, with a manual approver selection fallback if no flow applied.

A targeted version of our full inbox vision, focused strictly on high-signal attributes (deadlines and request types) while deferring complex domain-scoping filters.

Solved complex edge cases—like multi-quarter approvals—while keeping critical compliance safeguards intact for pilot validation.