Custom clinical operations software

MedPlexMD Injury (site), a Birmingham personal-injury medical practice, was paying every month for a third-party SaaS EHR they wanted to stop paying for. They had already started building their own portal to take over operations — but two pieces, secure video visits and e-prescribing, were keeping them tethered to the SaaS.

The work was not abstract “digital transformation.” It was concrete operational drag — staff switching between systems, manually reading dense medical documents, re-keying billing information, hunting for status across patient records, and trying to keep clinical work from falling through the cracks — compounded by the cost of running a clinic across two platforms.

Daedalus built the missing pieces inside MedPlexMD Injury’s own portal, then built the operational layer around them: a production AI document-processing pipeline, scheduling and telehealth wired into the same patient flow, role-aware clinical recommendations and billing review, and PHI-safe operational telemetry. The result was a clean migration off the SaaS onto software the practice owns — and can change in days when a workflow shifts, instead of waiting on a vendor roadmap.

At a glance

Documents extracted per month

335

Pages processed per month

1,955

Billing posts per month

110

Visit resolutions per month

94

Median successful lifecycle

2.2 min

Estimated staff time saved

60–100 hrs/mo

Average monthly run rate from PHI-safe Grafana / InfluxDB telemetry across the four complete production months from January through April 2026. Time savings use conservative manual-work assumptions (5–8 min per document, 0.75–1.5 min per page, 3–5 min per billing post).

The problem

MedPlexMD Injury was running a high-friction clinical operation where each step depended on accurate handoffs.

• Incoming referrals needed to become patient profiles, appointments, and tracked work.
• Telehealth was still happening through the SaaS. It needed to move into the portal and fit scheduling, patient status, call-room visibility, and billing.
• E-prescribing was still happening through the SaaS. Providers needed to launch it from the portal, with patient and provider data syncing cleanly into the prescribing system.
• Physician recommendations needed a safe prep and review workflow between case managers and providers.
• Records, imaging reports, prescriptions, ER records, and statements needed to be read, classified, assigned, and billed.
• Leadership needed a clear view of where patients and work were getting stuck.

“We’ve come leaps and bounds in a short amount of time. Daedalus helped fill the gaps so we could move off a traditional EMR onto a custom solution built for how we actually work.”

Bo Morgan, MedPlexMD Injury

That became the shape of the work: make the buried operational state visible, then automate the repetitive steps around it.

What we built

The two missing pieces that let MedPlexMD Injury leave the SaaS, plus the operational layer that turned the resulting portal into a clinic-wide system.

AI document processing

The pipeline continuously watches the portal’s document queue, marks new items as processing, converts PDFs into images, and uses a self-hosted vision model to extract structured data — document category, type, date, facility, providers, billing codes, medications, and other workflow-critical fields.

From there, a patient-assignment flow matches the patient, resolves the related visit, updates the document, creates notes, and posts billing fees. The system is instrumented with PHI-safe metrics that track every stage of the lifecycle without emitting names, DOBs, document IDs, or free-form PHI.

Portal workflow modernization

MedPlexMD Injury did not need a shiny sidecar app. It needed the actual portal to handle the work. We improved scheduling, appointment creation, intake, document queues, clinic summaries, billing review, user roles, and operational dashboards — keeping staff in one system and reducing the number of places a task could disappear.

Telehealth, owned instead of rented

Telehealth was the first of the two pieces still living in the SaaS. We replaced it with Zoom-backed virtual rooms using pooled host accounts, wired into the practice’s own portal — start and join actions tied to appointment and patient workflows, call history exposed, and the interface shaped around how clinic staff already moved patients through the day.

Recommendations, billing, and e-prescribing

Role-aware workflows let case managers prepare physician recommendations without making clinical orders, providers finalize them, and billing review outstanding work from a consolidated view. E-prescribing was the second piece still living in the SaaS — we connected provider setup, patient sync, and prescription launch flows so clinicians could prescribe from inside the portal without re-entering patient context.

How a document moves through the system

1. 01 Queue watch

   New documents are detected and marked as processing so staff can see the system has them.

2. 02 Vision extract

   PDFs become page images for a self-hosted vision model that classifies and structures content.

3. 03 Match patient

   The extracted data is matched to patient, facility, provider, and document metadata.

4. 04 Resolve work

   Visits, notes, medications, recommendations, and billing codes move through the portal flow.

5. 05 Measure

   PHI-safe metrics track extraction, matching, completion, billing, errors, and lifecycle time.

Where the monthly volume came from

The pipeline carries both high-volume simple documents and dense records packets. On a typical month, ER records and imaging dominate page count; imaging dominates document count.

Time saved

A conservative manual-work model puts the saved or redirected staff time at roughly sixty to one hundred hours per month.

That range is intentionally cautious. It does not count fewer missed handoffs, faster billing visibility, reduced interruption load, improved compliance posture, or the operational benefit of managers seeing bottlenecks directly.

Why it worked

Daedalus worked inside the mess instead of pretending it was cleaner than it was. We paired in live production-like workflows, watched staff drive the portal, used transcripts to preserve requirements, pushed small focused fixes, and instrumented the system so improvements could be measured.

The work moved from scheduling to telehealth to e-prescribing to document automation because those workflows are connected in the business — not because they fit neatly into a software module.

The result

MedPlexMD Injury now owns a complete operational platform: document processing, scheduling, telehealth, clinical handoffs, e-prescribing, billing review, and observability in one ecosystem — running on infrastructure they control instead of a SaaS EHR they were renting.

It is faster, more visible, and less dependent on manual stitching. And because the practice owns the code, a workflow change is a code change, shipped in days instead of a feature request waiting on a vendor roadmap. Most importantly, the system reflects how the clinic actually works.

Case study prepared from internal transcripts, repository review, and PHI-safe production telemetry. Patient information, private infrastructure, and the internal project codename intentionally omitted.