Promptly scales patient-care AI with DoiT’s Accelerator
Meet Promptly
Promptly is a fast-growing patient-engagement and workflow automation platform used by thousands of medical practices across the United States. Designed to streamline clinical operations, reduce administrative strain, and improve patient experience, Promptly helps practices manage communication, scheduling, digital intake, payments, reminders, and two-way messaging from a single connected hub.
Founded by CEO Dr. Anish Kapur, Promptly has become a critical part of the clinical workflow for practices where operational efficiency directly affects patient outcomes. As care models shift and AI becomes a strategic differentiator, Promptly is investing heavily in solutions that support automation, reduce staff burden, and enable better decision-making inside the clinic.
“We exist because medical practices need to move faster with fewer resources,” Kapur says. “The platform has to feel effortless for patients and powerful for providers. That balance is only possible with the right technical foundation.”
The Challenge
As Promptly scaled to serve more providers and patients, operational pressure on its engineering team increased sharply. More practices were adopting the platform, workloads were increasing, and customers asked for higher levels of automation, personalization, and intelligence. At the same time, Promptly was preparing to launch a new AI-driven feature designed to give practices a more predictive, proactive understanding of patient engagement patterns. This new feature would be launched as a premium add-on, meaning as it entered wider adoption, it would drive incremental recurring revenue and in turn, expand customer lifetime value across its practice base.
“This feature will open an entirely new revenue line for us,” Kapur says. “It’s a strategic product, not a technical experiment.”
The challenge wasn’t vision, it was whether the team could execute and scale fast enough to meet rising expectations amidst the growing complexity of the cloud environment underpinning Promptly’s growth.
Internally, the team juggled the multiple demands of rapid shipping, strict security requirements, data compliance, real-time communication workloads, and rising expectations from healthcare customers who rely on Promptly for mission-critical workflows. Adding a net-new AI feature meant introducing new pipelines, validation layers, and architectural complexity at a moment when infrastructure stability and delivery speed requirements remained business-critical.
“Our customers depend on Promptly for operational continuity,” Kapur says. “When you’re running thousands of workflows a day, unpredictability isn’t an option. We needed a partner who could help us build this new feature without slowing down momentum.”
Early AI prototypes showed promise, but turning POC into a production-ready capability required additional expertise. The capabilities of this AI feature are to turn raw healthcare interactions – appointment requests, patient messages, operational notes, and follow-up workflows – into structured, automated actions. The feature reads unstructured written text, interprets patient intent, classifies the request, and then routes it through the correct operational workflow without human intervention.
The volume and variety of data Promptly processes create significant challenges. . Every clinic, practice, and provider uses different language, formats, and internal rules. The AI must also understand medical context, detect urgency, and maintain compliance-grade accuracy. Building a system that can reliably interpret natural language, execute the correct workflow, and avoid errors in a healthcare setting created significant operational and technical complexity, particularly without an internal ML team dedicated to continuous model refinement.
“We had the idea,” Kapur says, “but we needed a way to make it real, reliable, and repeatable. That’s where DoiT came in.”
The Solution
When Promptly engaged DoiT, both teams aligned on a single goal: build a scalable AI capability that would power new features while strengthening Promptly’s long-term product foundation.
The engagement combined DoiT’s GenAI Accelerator, cloud architecture expertise, and hands-on engineering support into a collaborative build model designed to move fast and validate continuously to support real-world, clinical-ready features.
Establishing the technical foundation
The first step was designing a production-ready environment capable of supporting Promptly’s new AI feature with enterprise-grade reliability. DoiT worked with Promptly to define a secure model-execution pipeline, ensure protected data paths, and architect the supporting AWS services in a way that reduced operational overhead while increasing observability.
- The team focused on:
- Robust, secure request handling for patient-dependent workflows
- Reliable integration patterns with Promptly’s existing APIs
- Scalable model execution tuned for real-world performance expectations
- Explicit guardrails to reduce hallucination risk, prompt drift, and inconsistent output
- Monitoring and traceability to support safe deployment in healthcare settings
“From the first session, it was clear that DoiT understood both the technical and operational stakes,” Kapur says. “They weren’t just writing code, they were helping architect something that needed to survive in a clinical environment.”
Building and refining the new AI feature
The new AI capability required a sophisticated blend of model reasoning, contextual understanding, and consistency. DoiT’s Accelerator team worked directly with Promptly’s product and engineering leads to refine prompts, structure model outputs in predictable formats, and ensure the system behaved reliably across diverse datasets.
A critical hurdle was ensuring deterministic, machine-readable output formatting. Prior experiments led to unpredictable response structures, making downstream integration nearly impossible. DoiT solved this by implementing prompt-chaining and validation layers that produced uniform, machine-readable outputs on every execution.
“Consistency was everything,” Kapur says. “We couldn’t risk a scenario where one execution behaved perfectly and the next one broke the workflow. DoiT solved that.”
Collaborative development and rapid iteration
The working rhythm between Promptly and DoiT quickly evolved into a tight, high-velocity co-development loop. The teams used a shared Slack channel, real-time testing feedback, and iterative refinement based on real practice data. DoiT’s responsiveness became a critical asset.
“It felt like a true partnership,” Kapur says. “We sent questions or examples and got answers almost immediately. That pace is rare. It allowed us to move faster than we could have on our own.”
Funding and acceleration
AWS funding and DoiT’s GenAI Accelerator dramatically reduced the cost barrier of building the new feature. Without it, Promptly would have had to divert significant internal engineering resources or delay the launch.
“The Accelerator changed everything,” Kapur says. “It made this level of expertise accessible. Without it, the feature would have taken months longer or wouldn’t have happened at all.”
The Results
The impact of the collaboration was both immediate and foundational. Promptly moved from concept to a production-ready AI capability far faster than original internal timelines anticipated.
Faster time to value
By leveraging DoiT’s Accelerator, Promptly compressed what would typically be a 3-4 month internal build into a matter of weeks. This prevented major opportunity-cost delays across the product roadmap and freed the team to continue delivering on customer commitments.
“We saved months,” Kapur says. “And it wasn’t just time. It was the difference between building something experimental versus something customers can trust.”
Higher accuracy and consistency
The refined AI pipeline produced stable, structured results that outperformed Promptly’s early prototypes. The feature now generates reliable insights in a fraction of the time required by manual analysis, with deterministic outputs safe for production workflows.
“This is the first time we’ve had something that works every single time,” Kapur says. “DoiT helped us build confidence, not just functionality.”
Future operational impact
Promptly is also preparing to use the AI internally to support customer success teams tasked with retention. The feature will allow CSMs to generate personalized insights rapidly, reducing the need for manual data interpretation and improving customer engagement.
“We’re about to use the AI ourselves,” Kapur added. “Our teams will finally have the ability to pull insights instantly instead of spending hours analyzing data.”
What's Next?
Promptly and DoiT continue to work together on new AI capabilities that will extend the system beyond tactical recommendations into broader strategic insights. Future iterations may include creative analysis, anomaly detection, more advanced patient-workflow reasoning, and deeper cross-system pattern detection.
“Now that we have the foundation,” Kapur says, “we can start thinking bigger. DoiT helped us build something we can scale for years.”
Promptly plans to use its annual consulting relationship with DoiT to support ongoing AI innovation, relying on the partnership not just for technical execution but for product thinking and long-term architectural strategy.