Illustrative composite: How Meridian Health Cut Process Discovery Time Using AI-Powered Intake

The Challenge: Discovery Was the Bottleneck

Illustrative composite — figures: Meridian Health operates across 38 facilities and employs over 22,000 people. Their automation Center of Excellence, established in 2022, had made significant progress - 94 live automations across clinical operations, revenue cycle, and supply chain. But by mid-2025, the team had hit a wall.

Illustrative composite — figures: The problem wasn't development capacity. The team had developers available. The problem was that bringing a process from 'someone said we should look at this' to 'ready for development' took an average of six weeks - and the quality of what arrived at the development queue was inconsistent. Some processes had been thoroughly assessed. Others arrived with little more than a name and a vague description.

Illustrative composite — figures: 'We had fourteen people on the team, and at any given time, four or five of them were just doing intake work,' recalls Ana Rodriguez, Head of Healthcare IT at Meridian. 'Sitting in workshops, taking notes, building spreadsheets, chasing business owners for follow-up information. It was the least scalable part of our entire operation.'

The Root Cause: Intake Was a Manual, Artisanal Process

Illustrative composite — figures: The team's intake process relied on a combination of workshop interviews, a 40-question spreadsheet, and follow-up emails. Each intake required at least two in-person sessions with the business owner, one with the process SME, and a separate validation meeting with IT to confirm system access and API availability. The minimum cycle was four weeks. Complex processes - anything touching clinical systems - routinely stretched to eight or ten.

Illustrative composite — figures: Worse, the quality of the data that emerged from this process varied dramatically by the intake analyst assigned. Some analysts were rigorous about capturing volume metrics, exception rates, and compliance requirements. Others relied on estimates and rough descriptions. When these intakes hit the development queue, the discrepancy in data quality created wildly inconsistent effort estimates and, ultimately, unpredictable project outcomes.

The Approach: AI-Powered Conversational Intake

Illustrative composite — figures: Meridian piloted IntakeOS in September 2025, starting with their Revenue Cycle Management team - a department with high intake volume and well-understood process types. The pilot replaced the spreadsheet-and-workshop intake process with an AI-powered conversational interview that business users could complete asynchronously, on their own schedule.

Illustrative composite — figures: The AI interview - led by VARA, IntakeOS's AI business analyst - guided business users through a structured qualification process: problem statement, root cause analysis, pattern qualification, and backlog refinement. VARA adapted its questions based on user responses, probed vague answers, and captured structured data automatically. An intake that previously required two in-person workshops could be completed by the business user in 45 minutes.

The Results: More Pipeline, Better Quality

To apply the lesson rather than the composite's invented figures, evaluate IntakeOS on one healthcare workflow using approved, nonsensitive descriptions. Have frontline staff correct the process map and validate workload estimates before comparing opportunities. Keep clinical, privacy, security, and implementation review separate; this scenario does not establish an expected result for your organization.

Illustrative composite — figures: After 90 days, average discovery cycle time dropped from 6 weeks to 9 days. Intake analyst capacity freed by the automation was redeployed to solution design and quality assurance - activities that require human judgment and were previously under-resourced. Pipeline throughput tripled.

Illustrative composite — figures: But the quality improvement was equally significant. Because VARA applied the same structured qualification framework to every intake, the consistency of the data arriving at the development queue improved dramatically. Developers reported spending significantly less time in pre-development discovery. Mid-project scope changes - previously the team's most common source of delay - dropped by more than half.

"Illustrative composite — figures: VARA found things we wouldn't have found ourselves. It asked the root cause question directly - 'could fixing the underlying system solve this?' - and on three of our first ten intakes, the answer was yes. We would have built those automations and they would have been the wrong investments."

- Ana Rodriguez, Head of Healthcare IT, Meridian Health

Unexpected Discoveries: The System Fix Gate

Illustrative composite — figures: One of the most valuable outcomes of the pilot was unexpected. Of the first 24 processes submitted through IntakeOS, VARA's root cause analysis flagged three as system change scopes rather than automation candidates. Two involved Epic module configurations that the clinical team had not known were available; one involved a revenue cycle workflow that required a ServiceNow upgrade rather than an overlay automation.

Illustrative composite — figures: In the previous intake model, these processes would have progressed through development, been built as RPA automations, and eventually been either abandoned (when the underlying system caught up) or maintained indefinitely as technical debt. Catching them at intake saved an estimated 340 development hours and avoided three future maintenance burdens.

What's Next

Illustrative composite — figures: Meridian has now fully replaced its legacy intake process with IntakeOS across all three business units. The team has set a target of 200+ new automation candidates per quarter - more than four times their previous throughput - with no additional headcount. 'The bottleneck used to be discovery,' Ana notes. 'Now the bottleneck is development, which is exactly where it should be. That's a much better problem to have.'