How DentScribe Reduced Report Generation Time from 10-20 Minutes to 2 Minutes in Just One Week

In clinical settings, time is in short supply as a professional has to diagnose multiple patients in a day. But the actual problem starts when technology lags. Because it slows the process and leaves users frustrated. DentScribe was facing similar challenges. 

This case study shows how we helped them solve a key problem with their system. And the best part is that they could see the reliable performance in just one week.

Who Is This For?

• Dental practice owners and clinic managers seeking faster documentation solutions.
• Healthcare IT leaders needing real-world strategies to optimize clinical workflows.
• Software architects interested in best practices for scaling AI-powered healthcare tools.

What Problems Did DentScribe Experience?

DentScribe, based in Sunnyvale, CA, provides an AI-powered clinical document generation platform for dental professionals. Their tool enables clinicians to create SOAP notes and reports by speaking naturally, converting voice into structured, customized documentation—such as After Care Summaries, Perio Charts, and Specialist Reports.

As DentScribe’s adoption grew, the following pain points became critical:

Here are some problems they were facing:

1. Slow Report Generation Times: Creating a single report could take anywhere from 15 minutes to an hour. It disrupted clinical workflows and eroded trust in the system.
2. One-at-a-Time Queue Processing: Reports were processed one after another. If one report took a long time, everyone else waited. During busy periods, this created a backlog and frustrated users.
3. No Auto-Scaling or Dynamic Resource Allocation Capability: The backend didn’t adjust to demand. Whether there were 10 or 100 report requests, the system used the same single-threaded process, so delays piled up during traffic spikes.
4. Increased Load, No Visibility: With limited real-time monitoring, the team couldn’t see or react to usage spikes. They lacked the tools to scale resources when needed.
5. User Experience Was Suffering: These delays weren’t just technical—they left dentists and staff waiting for critical documentation, hurting productivity and satisfaction.

It was clear that DentScribe needed a smarter, more scalable system to keep up with its growing user base.

Why Is This a Common Problem?

As dental practices scale up their use of AI tools, report queue bottlenecks and lack of scaling become widespread issues. Clinics looking for the best AI dental documentation software need platforms that are not only accurate but also responsive and scalable during busy periods.

How Did ThinkSys Solve DentScribe’s Slow Report Generation?

The Proposed Solution: Answering Key Dental SaaS Bottlenecks

After a thorough analysis of DentScribe’s report generation workflow, ThinkSys recommended a targeted, multi-layered optimization plan. This approach addressed every major pain point and set DentScribe up for sustainable, user-focused growth.

Here’s How Each Problem Was Solved:

1. Parallel Processing with Multiple Workers: 
   • How do you eliminate slow, one-at-a-time report queues?
     By redesigning the system to process multiple reports in parallel, DentScribe could generate documents for several users at once. This significantly reduced wait times and improved clinician productivity.
2. Scheduled Scaling for Predictable Peaks: 
   • How do you handle busy clinic hours without overprovisioning?
     ThinkSys implemented scheduled scaling, automatically adding resources during known peak times (e.g., weekday mornings). This ensured consistent performance during rush periods while keeping costs under control.
3. Real-Time Auto-Scaling with CloudWatch Alerts: 
   • What about sudden, unpredictable traffic spikes?
     Real-time queue monitoring was set up using AWS CloudWatch. The system now automatically scales resources up or down based on actual demand—responding instantly to usage surges and keeping documentation turnaround fast.
4. Scaling Rules For Visible vs. Old Messages: 
   • How do you avoid both under- and over-provisioning?
     Scaling out is triggered by the count of visible messages in the queue, ensuring rapid response when demand builds. For scaling in, “old message” alerts ensure resources are only reduced after true traffic drops—preventing premature shutdowns.
5.  Efficient Resource Use During Off-Peak Hours: 
   • How do you keep costs down without sacrificing performance?
     Efficient Resource Use During Off-Peak HoursWith both scheduled and alert-based scaling, the system would intelligently balance performance and cost, scaling up when needed and scaling down when load drops.

This approach offered DentScribe a solid foundation that could grow with their business while keeping users happy.

Implementation: Step-by-Step for Maximum Impact

Step 1: Audited the Existing Queue-Based System: We began by digging into the report generation process under real-world load. Processing each report one at a time was the main source of delays. We mapped peak traffic, measured queue lengths, and pinpointed slowdowns. This gave us a clear view of where the system struggled and how resources were being used. This process gave us a clear overview to design a solution that would scale smoothly. 

Step 2: Designed a Multi-Worker Architecture: We rebuilt the system to let multiple workers process reports in parallel. Moving from single-threaded to concurrent processing cleared the backlog and sped up delivery. We addressed worker coordination, failure recovery, and shared resource access. To prevent jobs from clashing or duplicating, we fine-tuned job visibility and retries, especially with AWS SQS. 

Step 3: Set Up Scheduled Scaling Based on Usage Trends: By analyzing DentScribe’s usage logs, we identified high-traffic windows and scheduled extra resources for those times, like early weekday mornings. This handled predictable peaks efficiently and kept resource use in check. We set conservative limits so the system wouldn’t overreact to minor spikes. 

Step 4: Integrated Real-Time Auto-Scaling with AWS CloudWatch: To handle sudden surges, we set up step-based auto-scaling with AWS CloudWatch. The system checked the queue every 30 seconds. If the queue grew, new workers spun up automatically. We tested thresholds to avoid scaling too aggressively, which could drive up costs or strain the system. Robust error handling and rollback safeguards were built in. 

Step 5: Calibrated Scaling Logic with Old and Visible Message Alerts: We fine-tuned the scaling logic for both scaling up and scaling down. “Visible message” count triggered scale-out, so the system responded quickly to demand. For scaling in, we used “old message” alerts to avoid shutting down resources too soon. This kept the system lean when demand dropped, but ready for spikes. Live dashboards let us monitor and adjust scaling in real time.

Results: What Changed for DentScribe?

After launching the new system, DentScribe saw immediate improvements:

• Drastic Reduction in Report Generation Time: Reports that once took up to an hour now finish in just 1–2 minutes. This change made a real difference for users, who could move on with their day instead of waiting for documents to be processed.
• System Stability: The new architecture ran smoothly, with no downtime. Users could rely on the system, and the team no longer worried about unexpected outages.
• Faster Turnaround, Smoother Workflows: Reports were delivered quickly, reducing interruptions and keeping clinical work on track. Dentists and staff could focus on patients instead of paperwork.
• Increased Team Efficiency: Internal teams spent less time handling delays and support tickets, freeing them up for more valuable work.
• Consistent Performance During Peak Hours: Even at peak loads, DentScribe’s platform remained responsive, maintaining trust and satisfaction.

Why This Matters:

ThinkSys’s scalable architecture and proactive scaling strategy enabled DentScribe to deliver the fast, reliable documentation clinicians need—directly improving patient care and operational efficiency.

Conclusion 

DentScribe’s transformation shows how crucial it is to fix slow systems. By rethinking the architecture and scaling strategy, ThinkSys helped them deliver fast, reliable results that users noticed right away.