From 6-Month Backlog to 2-Week Turnaround: How Globex Financial Transformed Their CoE Pipeline

The Crisis Point

In Q3 2025, Globex Financial's Head of Automation, Claire Huang, presented the CoE's quarterly performance review to the executive team. The numbers told a difficult story. The automation backlog stood at 47 items, with an average intake cycle of six months. Of the 28 automations completed in the previous 12 months, 11 had been abandoned or significantly de-scoped during development - a 40% failure rate that was impossible to ignore.

'The business had stopped submitting ideas,' Claire recalls. 'They'd submitted things months ago and heard nothing back, or they'd had projects start and stall. The perception was that automation was slow, expensive, and unpredictable. We were losing the internal credibility we'd worked hard to build.'

Diagnosing the Problem

The root cause analysis pointed clearly to intake. Globex's intake process was a hybrid of workshop interviews and a 50-question Google Form that business owners completed asynchronously. The form data was inconsistent. The workshops were resource-intensive. And the qualification criteria - beyond a rough complexity/benefit matrix - were undefined.

More concerning was what the failure analysis revealed. Of the 11 projects that had failed or been significantly de-scoped, 8 had problems that were identifiable at intake if the right questions had been asked. Three involved processes where the underlying system could have been upgraded to eliminate the manual step - they should have been system change scopes, not automation projects. Two involved document processing use cases that required IDP, not RPA as originally scoped. Three involved processes with exception rates so high that automation ROI was questionable from the start.

The Transformation: Problem-First Intake with AI

Globex implemented IntakeOS in October 2025, replacing their Google Form and workshop intake process with AI-powered conversational intake. The change was phased - starting with the Compliance and Operations teams, which had the highest intake volume, before rolling out enterprise-wide.

The AI interview applied the Problem-First framework systematically: establishing the problem statement before discussing solutions, explicitly asking the root cause question before proceeding to automation qualification, walking through the seven-pattern decision tree to identify the right automation approach, and closing with the developer complexity assessment for backlog refinement. Each intake took the business owner 45–60 minutes and produced a structured qualification report that the CoE team could review without additional discovery work.

The Pattern Mismatch Discovery

One of the most valuable outcomes of the new intake process was the pattern mismatch analysis. Of the first 60 processes submitted through IntakeOS, 12 were flagged as system change scopes (where fixing or upgrading the underlying system was the correct recommendation), 8 were identified as API integration candidates (where direct integration was preferable to RPA), and 5 were classified as IDP use cases that had previously been scoped as RPA.

The re-classification of 25 candidates - more than 40% of the initial batch - was initially surprising to the team. 'We'd been solving some of these problems with the wrong tool for years,' Claire notes. 'The API integration ones are the most embarrassing in hindsight. We had ServiceNow-to-Salesforce data sync running as an RPA bot that required maintenance every time either system updated. That's a textbook case of RPA being applied where an integration should have been.'

"The root cause question changed everything for us. We'd been assuming for years that every submission was an automation opportunity. VARA taught us to ask first whether the system could be fixed instead. That question alone redirected millions of dollars of misdirected development."

- Claire Huang, Head of Automation, Globex Financial

Rebuilding Business Confidence

The most important outcome, in the long run, may be the rebuild of business confidence in the CoE. The combination of faster turnaround, more transparent qualification, and better project outcomes has reversed the attrition trend in intake submissions. By Q1 2026, Globex was receiving more new submissions per month than at any point in the CoE's history.

The CoE has now set a target of 100+ qualified candidates per quarter and is piloting a citizen developer program for low-complexity, departmentally-owned automations. The executive team has approved a 30% increase in the CoE's development capacity for 2026. The transformation that started with a new intake process has cascaded into a fundamental reset of the program's trajectory.