Charles Lyons-Burt
For over 20 years, Jason Adolf has assisted government agencies in identifying and implementing technologies that modernize their operations for today and the future. As the vice president of global public sector at Appian, he collaborates with defense and civilian agencies, as well as state and local governments and educational institutions to enhance and automate workflow processes using low-code application development, artificial intelligence and other innovative technologies.
Jason recently sat down with ExecutiveBiz for a Spotlight interview to discuss how AI is fueling advances in process innovation. He explained how AI not only makes current workflows more efficient but also provides agencies with an opportunity to fundamentally transform processes to encompass new capabilities and use cases.
ExecutiveBiz: Can you describe your role at Appian and some of your major responsibilities?
Jason Adolf: My role is to look after our global public sector vertical, and that involves a combination of several things. Part of my job is to set the strategy for how we approach the IT marketplace and evaluate new technologies as they become available. And then I also serve in an advisory role to many of our customers and partners on how to get the most out of their technology investments. So it’s kind of an intermediary role of connecting new capabilities in the marketplace with the emerging needs of our clients.
I started my career at SRA International and most of my roles since then have remained focused on large-scale systems integrations within government. My projects typically impact hundreds or even thousands of users and play a significant role in mission-critical operations. As a result, major public sector systems implementations have become second nature to me and have given me crucial insights about enterprise mission system transformations.
Many of our customers are navigating projects of that scope for the first time, investing in new technologies on an unprecedented scale and within a mission-essential framework they have had prior experience with. A lot of my role entails working with government customers to set expectations of what to expect when they’re investing in a modernization effort at this scale and offering support on the post-decision implementation and configuration issues that can occur.
EBiz: How are government organizations transforming their business operations with process innovation?
Adolf: Process innovation can be tricky because an agency may have a long history of doing things a certain way. Current processes may be well-established and now it’s time to innovate a different approach. As I mentioned, a modernization effort may be happening on a scale that an agency hasn’t encountered before. So there are a lot of unknowns in innovating at scale and moving the thinking beyond just process optimization to process innovation.
The distinction here is that process optimization is about improving the speed, accuracy, or efficiency of current processes. But modern AI technologies allow you to transform certain processes on a more fundamental level. So, for example, you could choose to optimize an existing process around managing disability claims by using AI to automate certain steps in that process. But it’s also possible to innovate on a more fundamental level by using AI to access more data in real time to evaluate and even help adjudicate those claims automatically.
Our role is to help agencies reconceive what’s possible with AI and help them think more creatively about how to modernize their processes to support these new possibilities. It’s important to note that process innovation involves a lot of different levels of the organization — from technology leaders and provisioning officers to system operators, case workers, grants managers and others who all have a role to play in how that particular process plays out in the agency.
EBiz: How can the public sector balance responsible AI adoption without hindering innovation?
Adolf: Compared to the private sector, government agencies tend to operate in a more highly regulated environment. This is true both in terms of the rules governing agency operations and in terms of the nature of the data itself, which can include PII of citizens or even sensitive classified information.
Against this backdrop, we’re now seeing more and more policies around the use of AI, specifically the use of public AI services.
What this means is that trusting in the performance and outputs of AI is becoming more than just a quality assurance issue, but also now a matter of compliance for government agencies. And the stakes for this are often higher in government than in the private sector. It’s one thing for AI to mix up an order or inadvertently share customer data on an ecommerce website; but if this is happening in a FEMA program for disaster relief or a DOD program tied to national security, the implications may be far more serious.
This is why we’re seeing more agency teams employing retrieval-augmented generation, or RAG, prompt engineering and other steps aimed at enhancing the accuracy and relevance of AI outputs. Agencies are also turning more often to private AI deployments to ensure confidentiality around data and modeling. Rather than using public infrastructure, private AI brings algorithmic processes in-house, so you can conduct modeling and train on data that is not shared outside the organization.
EBiz: Can you share a specific instance or use case for process innovation in government?
Adolf: A good example would be the use case of processing disability claims that I mentioned earlier — where we don’t just want to speed up the current process, but actually use AI and process intelligence to handle some of the more basic claims autonomously, up to and including adjudication. Successfully innovating at this level requires us to solve multiple technology challenges throughout the steps of a process.
For instance, with this use case, you will want to employ AI and process intelligence that can automatically have authentication tie-ins to a person’s medical files and other relevant databases; and we need to make sure access is properly encrypted so that it doesn’t need to go through an analyst to approve access. Further along in the process, when the system is actually adjudicating and making a decision on the claim, we need to ensure traceability to understand the AI’s chain of logic and ensure the decision is being made correctly.
Throughout, we want to continually refine our models so that we start with the easiest of decisions — things that are relatively black and white — and then train the system over time to handle more nuanced or complicated claims.
EBiz: What are some of the key challenges or issues agencies are facing going into 2025?
Adolf: Top of mind is to continue to build confidence in autonomous decision-making within systems. Especially as use cases become more mission-critical, whether that’s a VA disability claim or a DOD tactical system, we’ll be looking to build more confidence around how those autonomous decisions are made and allowing something other than a human to make an actual adjudication. Automated decisions should be fast, but also accurate enough so that cases of appeals, where review cycles are added to scrutinize and potentially revise an automated decision, remain rare.
More generally, the other learning curve I see in 2025 involves how agencies can better support what I would call large, long-term continual optimization projects from both a technology and a provisioning perspective. Traditionally, agencies procure technology with some kind of base year, followed by a number of option years of ongoing maintenance. What I’m working toward is for agencies to embrace more low-code and related technologies in the market that are designed around continual improvement, and to help the federal government catch up to that as a contracting model.
I often say that a modern low-code technology deployment should never become legacy; it’s always improving. And so this combination of a continual improvement mindset backed up with federal contracting models that recognize and accommodate this paradigm is what’s needed for process innovation to really flourish.
