In the ever-evolving landscape of healthcare technology, efficient data management and compliance with stringent regulations are paramount. Internal tools play a crucial role in the HealthTech industry by ensuring data security, optimizing critical workflows, and driving operational excellence.
Airplane's Founder and CEO, Josh Ma, spoke with Robert Coe, Co-Founder and CTO of AcuityMD, and Abel Salcido, Deputy CTO of Electronic Caregiver (ECG) on how healthcare companies can maximize operational efficiency without compromising security with robust internal tooling. You can listen to the conversation or else read the transcript below. Enjoy!
Josh Ma: I'm Josh, the CEO and founder of Airplane. We are a developer platform for building internal UI and workload automation, all in one place. Many of our customers use Airplane to enhance operational efficiency in sectors like SaaS, fintech, healthcare, and healthtech. Today, we'll discuss internal tooling in the health tech space with Abel from Electronic Caregiver (ECG) and Robert from AcuityMD. Thanks for joining us, and let's dive into this topic.
Abel and Robert, could you please introduce yourselves and your companies? Abel, would you like to start?
Abel Salcido: Good morning, everyone. I'm Abel Salcido, the Deputy Chief Technology Officer for Electronic Caregiver. My background is in mechanical engineering with a keen interest in machine learning. Electronic Caregiver provides advanced telecare services, including technology and premier services along with peripheral devices to monitor your health throughout the day.
Robert Coe: Hi, I'm Robert Coe, the CTO and co-founder of AcuityMD. My background is in enterprise SaaS. AcuityMD is a commercial platform serving the medical device industry.
Josh Ma: Great to have both of you here. Could you share what your teams are currently working on and any top priorities? Additionally, what’s the most surprising thing you've seen in the intersection of technology and healthcare?
Robert Coe: At AcuityMD, we're excited about building a world-class data product that helps medical device commercial teams bring novel medical technology to patients. The medical technology field is advancing rapidly, but it's challenging to navigate the healthcare industry to get these innovations into the hands of doctors and patients. We're working on CRM functionality tailored to the medical device industry, contract management, and data products for analyzing patient journeys and their impact on go-to-market strategies.
One of the most surprising aspects of entering the healthcare space is the sheer breadth and depth of complexity within it. Each area of healthcare has its own nuances and complexities, making it a fascinating but intricate field to work in.
Josh Ma: Have you found specific niches or sub-verticals within the medical device industry that you're focusing on?
Robert Coe: Yes, we're focused on areas like case management and orthopedic implants. These niches have unique challenges that require specialized solutions, such as mobile usage and offline functionality for healthcare professionals who are often on the move.
Josh Ma: Abel, how about your team at ECG?
Abel Salcido: We're currently developing our proprietary virtual caregiver, a smart home health system that guides individuals through their daily healthcare routines. This includes medication management and vital sign monitoring. Our platform also collects and analyzes data to provide valuable insights.
The most exciting aspect is the potential to use this data to augment the work of doctors and nurses, especially in addressing the shortage of healthcare professionals. It's a promising direction, although some individuals remain skeptical due to concerns about data and AI. We hope to alleviate these concerns with ongoing discoveries and documentation.
Josh Ma: Let's delve deeper into the topic at hand and start by defining operational efficiency. How do you each define it in your businesses, and how is it measured or defined? Also, how does operational efficiency relate to your business KPIs, and who is responsible for it within your organizations? Abel, would you like to start?
Abel Salcido: Operational efficiency for us encompasses all our operations within the team. We believe in a collaborative approach, where everyone contributes to the end business goals. We hire team members for their expertise and allow them to bring their ideas to the table. This helps us maintain a welcoming environment and fosters team growth. Operational efficiency is owned by all teams within our organization, and everyone is encouraged to participate in the decision-making process.
Josh Ma: How do you assess your team's performance in terms of operational efficiency? What signs indicate success or failure?
Abel Salcido: Success for us is defined by meeting the goals set by our stakeholders. We work with product owners who create detailed stories, and when we successfully deliver a product that meets all the specified criteria, that's a successful outcome.
Josh Ma: Robert, how does AcuityMD approach operational efficiency and its measurement?
Robert Coe: At AcuityMD, we're transitioning from a startup to a scaling business, and operational efficiency is crucial. We define operational efficiency differently for each team, depending on their specific responsibilities and tasks. We measure efficiency by tracking factors like the hours spent on tasks, customer support effectiveness, and developer happiness. The goal is to make our teams more effective and satisfied in their roles. When operational efficiency is high, it leads to better products and faster iteration cycles.
Josh Ma: Can you share how operational efficiency connects to your business's ultimate goals and customer value?
Robert Coe: Operational efficiency ties closely to customer value. If our teams are more efficient and satisfied, they deliver better products and improve the customer experience. For example, in our enterprise software space, the speed of implementation is critical for a positive customer journey. We aim to make the onboarding process smooth and efficient. By doing so, we reduce friction for our customers and enhance their experience. Speeding up implementation is a fundamental aspect of delivering value to our customers.
Josh Ma: It's clear that operational efficiency not only impacts team morale but also plays a significant role in customer satisfaction and business success. Thank you both for sharing your insights.
The trade-off between operational efficiency and security and compliance is a significant challenge, especially in regulated industries like healthtech. Robert, from your experience, how has this transition been from a non-health tech background to a more regulated industry, and how has it affected your teams?
Robert Coe: Transitioning to the health tech space from a non-health tech background has brought more attention to security and compliance. The sensitivity of the data and the heightened concerns around security and compliance in the health tech industry require a strong focus on meeting standards from the very beginning, including preparing for audits like SOC 2. It's easier to establish these controls early on rather than later.
Josh Ma: Security and compliance measures can be resource-intensive. Do you find that your teams have the necessary access to systems to perform their jobs, or does this become a hindrance in the process?
Robert Coe: Access control and security measures are crucial in a regulated industry like health tech. We have to be diligent from the early stages in setting up single sign-on, employee group synchronization, and access controls to ensure that only authorized personnel can access sensitive data or tools. It's challenging, but it's a necessary investment in our industry.
Josh Ma: Abel, in the health tech space, how have you observed the evolution of security and compliance regulations, and how have they impacted your company?
Abel Salcido: Security and compliance regulations have become stricter, especially as we expand internationally. European regulations, for example, impose strict data localization requirements. It adds complexity and often results in monetary and resource impacts. We need to consider compliance with various regulations and often choose to follow the strictest requirements.
Josh Ma: How does this impact your product development process? Does it slow down the team's ability to build and iterate on products?
Abel Salcido: Documentation and regulatory approvals are the primary areas that slow us down. Compliance is a continuous process, and we have to keep monitoring and adapting to changes in regulations. The main bottleneck is getting clearance and proper documentation to start using new products or features.
Josh Ma: Given these challenges, how do you decide where to invest in tools or processes to support your teams and counteract slowdowns caused by security and compliance requirements?
Abel Salcido: We are a small development team, and we prioritize our investments based on the value we can bring to our stakeholders. When it comes to build versus buy decisions or implementing new tools, we often opt for third-party solutions if they align with our core competencies. Our primary investments are focused on internal platforms, helping us maintain a leadership position in the space. We also have to be HIPAA compliant. So, any service we use, our first question is, "Are you HIPAA compliant? Can we have a Business Associate Agreement (BAA) and necessary agreements in place to do business?" If these requirements aren't met, we usually don't proceed with the organization.
Josh Ma: That makes sense. What about you, Robert? What is your framework for prioritizing investments?
Robert Coe: Our general framework prioritizes customer value first. We face the challenge of deciding where to allocate our resources, especially in balancing internal use cases versus delivering direct value. It's essential for our teams to run the business efficiently. We focus on what we can automate, recurring tasks, and processes that people are managing in spreadsheets. We work with various stakeholders to identify pain points and make the teams more effective. It's a continuous process.
We have a two-pronged approach for this as well. Firstly, we regularly review pain points with customer success, sales, consulting, and support teams to identify slow areas that need improvement. These are prioritized from a product perspective. Secondly, I have personal check-ins with other stakeholders to understand their pain points and identify quick wins that don't require extensive product scoping.
Josh Ma: In your experience at AcuityMD, how often are fixes small versus major efforts? If a major effort is required, what does that usually look like?
Robert Coe: We aim to be agile and work incrementally. We start with small MVPs to determine if they genuinely help. These small changes accumulate over time and develop into comprehensive tooling suites that significantly enhance the team's effectiveness. We prioritize agility and strive for self-service on the engineering side. Self-service allows engineers to automate tasks, sometimes even on an individual level. If someone builds a script to solve a common issue, it can quickly benefit the entire team.
Josh Ma: So there's a complementary aspect of being self-serve and an agile approach to building your tools. It's not about massive monolithic projects but addressing problems as they arise and evolving solutions over time. Is that correct?
Robert Coe: Exactly, you've summarized it well. Self-serve capabilities and an agile approach allow us to solve problems quickly and effectively as they arise.
Josh Ma: It's clear that making informed decisions and focusing investments on core competencies can help teams balance the challenges posed by security and compliance regulations. Thank you both for sharing your insights on this complex topic.
It's fascinating to explore the journey of internal tooling within your respective organizations. Let's delve deeper into this process. Robert, could you share more about AcuityMD's journey with internal tooling and how it has adapted to the company's growth?
Robert Coe: At AcuityMD, we began with a homegrown tool that used open-source components. While it had certain advantages, such as using a familiar tech stack and the same development team, we soon realized the limitations. The open-source internal tooling lacked essential features for scalability, permissions, and access controls, which are crucial as the company expands. We also found that the scripts and runbooks became increasingly complex, requiring extensive institutional knowledge that could become a burden as the team grew.
Josh Ma: Robert, when you mention scaling, could you elaborate on the different dimensions that became challenging as the company expanded? What were the major factors that led to the decision to adopt a platform like Airplane?
Robert Coe: Absolutely. As a company grows, multiple dimensions come into play. These include factors like the increase in headcount, the growing size of the customer base, and the expansion of product surface areas. Each of these dimensions contributed to our need for improved internal tooling. The tipping point for us was the realization that we needed enhanced security measures and the awareness that our existing tooling was not keeping pace with our scaling needs. The gradual accumulation of scaling challenges made it clear that we required a solution like Airplane, which simplifies tooling and enhances the user experience. This shift aligned perfectly with the efficiency standards we set for internal tools and allowed us to overcome our scaling challenges more effectively.
Josh Ma: Abel, similarly, I'm curious about the history of internal tooling at ECG, maybe way back from the beginning. What did the early days look like? And how has that evolved?
Abel Salcido: It was interesting since we had two separate teams, both with around 20 engineers per team. One group was dedicated to internal tooling, and that's all they did. The other group was mostly focused on new developments. But as we started growing and scaling, and we scaled a lot faster than what we expected, we weren't able to keep up with new features and our legacy internal tooling.
So we wanted to start pivoting some of the engineers on our internal tooling team so we could meet customer demand. We were looking for a tool that allowed us to automate much of the scripting we had and democratize access to our data. We didn't want to give direct access to our tables, especially for our business-side team members who might not be familiar with SQL. We wanted a tool that made it user-friendly and quick for them so they could focus on their needs while we continued development.
Josh Ma: You mentioned that in some aspects the team wasn't able to keep up with the tooling for the new products you were launching. What did that look like? Were there incidents or near incidents that happened as a result?
Abel Salcido: I think Robert touched on it. The challenge was keeping up with the internal tooling, specifically the libraries. We'd release a tool from the internal side, and it would work well for about a year. Then, suddenly, we'd start encountering bugs because of library changes, etc. We had to pull the original engineer who developed it back into the tool, slowing down progress on the front end.
It made for a difficult balance between these tasks. Engineers prefer focusing on their projects to complete them, and if you constantly pull them back and forth, it can create a challenging work experience.
Josh Ma: You mentioned your internal tooling team was quite small, and you've reduced it. Can you share what these engineers are working on now, and how you've allocated their efforts?
Abel Salcido: We went from 20 engineers to just 2 engineers handling internal tooling. Today, those engineers still handle some tickets. They monitor the tools to ensure they're working correctly, and if someone reports an issue, they troubleshoot it. However, most of their time is spent on new development. About two hours a day are dedicated to the internal tooling that we use with Airplane, while the rest of their time is usually focused on modern applications and connections.
Josh Ma: Cool, that's a great segue into the next question. I'm curious to share with the audience, one or two examples of use cases where Airplane has been helpful. What are some concrete examples, and how have you seen the most significant leverage from Airplane?
Robert Coe: Sure, we've leveraged Airplane to build a data curation interface, allowing non-technical users to make edits to tables without needing to write SQL queries. Another use case involves empowering our data product team, which typically lacks frontend development experience, to create runbooks with user-friendly interfaces. Airplane's developer console has been especially helpful for team members who aren't well-versed in frontend development but want to build user interfaces for data workflows.
Josh Ma: And at a high level, what kinds of runbooks or tools might a data team be building?
Robert Coe: These runbooks often cover tasks like data deployments and quality checks, particularly for moving data across environments. Some of these are automated workflows, while others require human involvement for input and decision-making.
Josh Ma: I see, so it's a mixture of both automated and human-involved tasks. How do you envision this evolving over time, especially for both technical and non-technical teams?
Robert Coe: We aim to enable more engineers, particularly those with a semi-technical background, to own their tooling for specific use cases. This autonomy empowers them to solve problems without depending on a different backlog.
Josh Ma: And Abel, curious about your team's usage of Airplane. Can you share the top two use cases that come to mind?
Abel Salcido: Certainly, there are two significant use cases. The first one involved migrating 15,000 subscribers from our legacy platform to a newer one within a tight deadline. Airplane allowed non-technical team members to trigger the migration, making it a seamless process even during continuous 24/7 operations. The second use case resulted in significant cost savings by automating subscriber cancellations for non-reimbursable services, which saved us approximately $600,000 annually. Our customer support team was able to run this themselves without needing engineers to intervene.
Josh Ma: That’s great. It’s awesome to see how customers build in Airplane and it always is more impressive that we could’ve imagined. How do you both see your tools evolving and scaling with your business in the coming years? Robert, let’s start with you.
Robert Coe: We see the ease of use and self-service aspects of the tools expanding. We aim to empower more non-technical team members, such as those in professional services or implementation, to write SQL and perform basic coding tasks. Additionally, we're excited about the AI capabilities of Airplane, particularly the Autopilot function that can generate code automatically. We're interested in making it easier for less technical team members to self-serve and handle tools more effectively.
Josh Ma: Abel, do you agree or disagree with Robert's perspective on AI and its impact? Anything different you'd like to add?
Abel Salcido: I agree with Robert. The AI features, like Autopilot, will be beneficial because non-technical team members will be able to use them without needing engineers to be directly involved. For internal tooling, we anticipate continuous scaling as we onboard new organizations, each with their unique needs. Giving non-technical users access and training on running scripts will continue to be a necessity for ECG.
Josh Ma: Concerning the use of AI in tool creation, how do you ensure responsible use and prevent AI from causing issues in production?
Abel Salcido: At the end of the day, someone will oversee and review the tools generated by AI, such as our engineering support team. They'll validate that everything is functioning correctly and that access is appropriately restricted. The process will involve reviews and responsible oversight.
Josh Ma: So, it's not fundamentally different from any other engineering work. There's a review process and access control. Robert, what do you think about this?
Robert Coe: It's like an advanced auto-complete with AI assistance, and while we leverage GitHub Copilot for engineers who want to use it, it still goes through the same review and oversight processes.
Josh Ma: Great, that makes sense. It's an interesting and exciting development. It's fascinating to hear how you're using Airplane and your insights into the future. Thank you for sharing your experiences and thoughts. We’ll now open it up for Q&A from the audience.
Someone asked, how is your organization thinking about balancing security concerns and the use of AI language models?
This is a great question. It’s referring to various use cases where you're working with AI. In some cases, data might be sent to external data providers or model providers. In other cases, automated decisions are expected. And there's a third case where sensitive context is provided, and it's not expected to divulge that context. It's a complex issue, and while it's early days, people are still figuring things out. I'm curious about what you've learned so far. Let’s start with Robert.
Robert Coe: The concerns you highlighted are valid. Balancing security and AI, especially when dealing with internal documentation and chatbots, presents challenges. Sensitive information exists in virtually all businesses. Telling AI not to release private information is challenging, if not impossible, given the current capabilities. Restricting access to read-only, without allowing it to write or alter data, is one approach to manage risks. These are key areas to address.
Josh Ma: Abel, what are your thoughts on this?
Abel Salcido: Limiting the scope of access is another important step. Validating every process before deployment and thorough testing are essential in mitigating security risks. While this might not guarantee perfection, it helps minimize security breaches.
Robert Coe: To add to that, sending data to third parties is another critical aspect, and self-hosting becomes a crucial consideration as you move beyond the minimum viable product stage. We are experimenting with different vendors and approaches, but the rapid pace of AI development requires constant review and adaptation.
Josh Ma: What about you, Abel? Are you also running any AI models on-device, or is it all in the cloud?
Abel Salcido: We currently run all our models in the cloud. We did experiment with on-device models in the past, but due to latency issues and user-side complications, we discontinued that approach.
Josh Ma: Thank you for that question! We have one more.
How has AI improved your team's efficiency so far? If it has or hasn’t, what's the trend line like? Robert, you mentioned Copilot. Are you seeing dramatic changes?
Robert Coe: That's a good question. Measuring developer efficiency is notoriously difficult. I'm mostly focused on the happiness quotient and empowering developers to use any tools at their disposal. Instead of getting into the details of how much code they're writing or the number of pull requests, I'm focused on those who want to use AI effectively and are happy doing so.
Josh Ma: Is it mostly developers or are non-engineers using AI more as well?
Robert Coe: Our entire team is using AI for various use cases, including writing code. It's been fantastic for marketing, sales, and other departments as well.
Josh Ma: Abel, how about your organization? Has it been pretty widespread or is it different?
Abel Salcido: It's widespread internally within our development group. I haven't seen it used much in other groups within ECG. It has been helpful when our engineers face blockers and use AI to find solutions. It has helped at times, although it can sometimes lead to more questions.
Josh Ma: Well, awesome, folks. This is about all the time we have today. We can wrap up here. Abel, Robert, thank you so much for your time. I've learned a lot, and I hope the audience has too. I wish you the best of luck in your journey to improve operational efficiency and in building your businesses. Thank you!
Abel Salcido: Thanks, Josh.
Robert Coe: Thank you, Josh. Bye, everyone.