Every technology has a maturation cycle; today we see Artificial Intelligence transitioning from being a parlor trick to being considered for serious applications. The federal government wants secure and reliable solutions to solve problems in the military and healthcare.

Our guest today is Dr. Ellison Anne Williams, she has a PhD in mathematics and is the founder of Enveil. She provides an overview of AI security by suggesting it is only as good as the data over which you train and use it.

AI is exposed to large data sets and models are encoded with the data with which they were trained. This process can leave the model vulnerable and open to attack, she describes one attack called a “model inversion.”

This is a machine learning technique that examines a model’s output and infers personal information about its data subject.

Dr. Ellison suggests a group of technologies called “Privacy Enhancing Technology.” During the interview, she gives an overview of how it can securely and privately train a model to produce richer insights.

PET allows leaders to secure the use of a wider range of data sources. You can use homomorphic encryption to safely train your model over sensitive data.

This interview is an overview of a technology that can allow federal agencies that must deal with sensitive information to be able to leverage the speed and insights that AI can provide. 

Want to leverage you next podcast appearance?  Take the quiz. https://content.leadquizzes.com/lp/fk1JL_FgeQ

Connect to John Gilroy on LinkedIn   https://www.linkedin.com/in/john-gilroy/

Want to listen to other episodes? www.Federaltechpodcast.com

Years ago, people would laboriously code character by character. This tedious process would take hours and would include errors. Over the years, libraries of prewritten code have evolved that allow software developers to “grab” some code, modify it, and finish a project earlier.

Malicious actors have taken advantage of this short cut and have injected code into these software libraries that get taken along for the ride.

One proposed solution is something borrowed from the shipping industry. A commercial invoice may be packaged with a bill of lading to indicate the contents of the package. This “assurance” has been transferred to the world of pre-written code and is now called a “Software Bill of Materials,” or SBOM.

In a world where you are shipping a ton of Portland Type II cement overseas, this bill of lading works finds; it has some challenges being transferred to the dynamic world of software.

In a typical federal environment, there is continuous change in the code itself. It would be difficult to change on ton of a manufactured product like Portland Type II Cement. However, the once approved software package may have so many changes that the Software Bill of Materials may not have any validity.

During the interview today, David Jurkiewicz unpacks the concept of an initial SBOM and then how software packages can evolve over time and still retain compliance. His company can take this basic guarantee and examine the software for many concerns, including.

·       Vulnerabilities

·       Dependencies

·       Integrity

·       Malware

·       Foreign presence

·       License

David Jurkiewicz provides details on how companies can resolve vulnerabilities and ensure safe operations in a world where code is grabbed off the shelf and slipped into a package.

Want to leverage you next podcast appearance? https://content.leadquizzes.com/lp/fk1JL_FgeQ

Connect to John Gilroy on LinkedIn   https://www.linkedin.com/in/john-gilroy/

Want to listen to other episodes? www.Federaltechpodcast.com

One of the most practical applications of Artificial Intelligence (AI) is to assist in network observability. The big move to Zero Trust is predicated on the ability to have a thorough understanding of network assets.

This is a significant issue for federal information technology. We have legacy systems, shadow IT, and a deluge of data in addition to the confusion that a hybrid network can bring.

Riverbed takes a phrase from science, telemetry. Initially, it was used to troubleshoot the original network: the power grid. Since then, the term has been modified to apply to a standard data collection system for analyzing information on a digital network.

The fact that 98% of the Fortune 100 uses Riverbed for determining network status means that they are the de facto leaders in the market.

Today, we sat down with Jeff Waters to help us understand how Riverbed can be applied to federal systems. You would expect Jeff to emphasize network management, however, he shows how the basic “telemetry” approach can be used for improving user experience.

The approach is simple:  if a technology can look at movement on a network, it can be applied to understanding how federal sites are used by citizens.

We move from DevOps to Artificial Intelligence Ops, or AI Ops at the end of the interview. This concept allows Riverbed to be able to understand a situation and offer remediation. Because the network is so well understood, the solution is effectuated quickly.

Telemetry – from old-school electrical troubleshooting to helping with user experience on a federal website.

= = =

Want to leverage you next podcast appearance? https://content.leadquizzes.com/lp/fk1JL_FgeQ

Connect to John Gilroy on LinkedIn   https://www.linkedin.com/in/john-gilroy/

Want to listen to other episodes? www.Federaltechpodcast.com

In 1848 they found gold at Sutter’s Mill in California. According to the CEO of Cohesity, data is the new gold. Today, we sit down with David Kushner from Cohesity to unpack what this “gold” reference means for federal technology leaders.

This has always been the case. Today, we see cheap storage, fast Internet, and Generative AI producing an overwhelming amount of data sets. The challenge is how to protect them.

This “perfect storm” has not gone unnoticed by federal leaders. If you casually look at a few recent federal mandates, you constantly see a reference to “security” and Artificial Intelligence.

·       White House: President Biden issues Executive Order on Safe, Secure, and Trustworthy Artificial Intelligence

·       Homeland Security:  Promoting AI Safety and Security

·       OMB:  Advancing Governance, Innovation, and Risk Management for Agency Use of Artificial Intelligence

Cohesity has been in business for over a decade and has garnered a reputation for efficient management of critical assets. We sit down with David Kushner to unpack what these innovations mean for the federal government.

For example, we all know that the federal government gets attacked thousands of times a day. We all know the “usual suspects,” but what is starting to happen is malicious code is being injected into backups. It is conceivable that a systems manager could reach for the backups and introduce compromised data into a sensitive data set.

During the discussion, David mentions that Cohesity has worked with over three hundred federal agencies in a wide variety of services. In February of 2024, Cohesity launched a search assistant called Gaia. This allows enterprise-level organizations to use Large Language Models and Retrieval Augmented Generation in a manner that complies with compliance mandates.

Listen to learn about Gais, backups, and a new world where the data is the gold that is being exfiltrated.

= = =

Want to leverage you next podcast appearance? https://content.leadquizzes.com/lp/fk1JL_FgeQ

Connect to John Gilroy on LinkedIn   https://www.linkedin.com/in/john-gilroy/

Want to listen to other episodes? www.Federaltechpodcast.com

We are at the point where Artificial Intelligence is splitting into several capabilities. These include Generative, Natural Language Processing, Computer Vision, Predictive, and Deep Learning to name a few.

Today, we sit down with Melisa Bardhi from Excella and look at one aspect of AI, Generative AI, and examine how it can be used responsibly for federal applications.

In a short 25 minutes, she covers

·       Definitions

·       Beginnings

·       Security

·       Existing models

·       AI training

Melisa begins by admitting that one must be cautious about data that is used to feed AI.  All humans have biases, whether intentional or not. One document she would like to share with the audience is Excella’s Generative AI Use Policy. 

Rather than starting with a major tech overhaul, it is suggested that an agency looks at a pilot with well-defined outcomes.  Generally, Generative AI should be deployed in a controlled environment.

Security is foremost in the minds of all federal employees. The challenge is that many of the recommendations from laudable organizations like NIST can be hard to understand.  As a result, Excella has unpacked many security concepts in several blogs. A reference is Decoding Artificial Intelligence: A Simplified Guide to Key Terminology.

  https://www.excella.com/insights/decoding-artificial-intelligence-a-simplified-guide-to-key-terminology

When you read about ChatGPT producing summaries, this is an example of a model that is used in a wide-open environment.  Melisa suggests agencies test out pre-trained models through cloud platforms like Azure ML, AWS Sagemaker, and GCP AI Platform.

Fortunately, we are bombarded with places to learn more about Generative AI.  During the interview, Melisa mentioned agencies should consider partnering with research institutions and groups like ACT-IAC. For example, you can start your Generative AI journey with an article titled “AI Understanding in the Federal Government.” 

https://www.actiac.org/documents/ai-understanding-us-federal-government

= = =

Want to leverage you next podcast appearance? https://content.leadquizzes.com/lp/fk1JL_FgeQ

Connect to John Gilroy on LinkedIn   https://www.linkedin.com/in/john-gilroy/

Want to listen to other episodes? www.Federaltechpodcast.com

Sometimes, we must look at literature to get an understanding of the current situation for federal GWAC contractors. Let us look back to Charles Dickens with, “It was the best of times, it was the worst of times.”

Here is the scenario: your company spends a ton of time and money to get a spot on a GWAC contract. never to win any work. All your hopes and dreams of increasing business have been crushed.

Today, we sit down with Brian Seagraves from Deep Point Water & Associates to take a detached view of this GWAC process and provide some suggestions.

We begin with some shocking numbers. We all know that federal organizations working under GWAC send out notifications regularly. In a perfect world, a resolute employee shows up for work at 9 AM, sees a notification, and then begins the filtering process.

In the real world, companies are overwhelmed by notifications and end up working to dead ends.

For example, one organization gave 27,450 announcements annually. Simple math indicates, that with 250 workdays, this is 100+ notifications a day. Mere humans cannot keep up. Inevitably, their companies may pursue bad fits or even miss excellent opportunities.

During the interview, Brian suggested a solution from Deep Water Point & Associates that can assist in this complex filtering work. His company has managed to combine a strong knowledge of keywords and human understanding of culture to assist in the selection of which announcements to pursue.

He details how they developed GWAC NorthStar. This is a “as-a-service” offering and a company can be set up with a free 30-day trial.

GWAC NorthStar is a good example of how Artificial Intelligence can be combined with human “know-how” to speed up the proposal process.

= = =

Want to leverage you next podcast appearance? https://content.leadquizzes.com/lp/fk1JL_FgeQ

Connect to John Gilroy on LinkedIn   https://www.linkedin.com/in/john-gilroy/

Want to listen to other episodes? www.Federaltechpodcast.com

We are in a whirlwind of change. Data is pouring into federal agencies.

A recent report indicated that data was growing at 25% a year; cloud usage at a rate of 61%; and Software as A Service an astounding 200%. Oh, did I forget to mention that Generative AI is also increasing data exploding?

It is one thing to worry about data from John’s Doughnuts, and quite another to worry about sensitive military, financial, and health information that the federal government is charged to protect.

Data. gov estimates that 250 million data sets are being used by the public sector.

This increase in data has not been missed by the White House. On February 28, 2024, an Executive Order called “Protecting America’s Sensitive Data” was issued.

Today, we sat down with Adam Costello to have him share some of his practical knowledge of securing data. Adam begins by suggesting federal agencies start by figuring out where these data sets are stored. Humans make mistakes, humans jump into shadow IT and create ghost data, and people do not understand retention protocols.

Adam suggests that an Asset Management Database can establish a baseline for documenting data. From there, we can get a better perspective on concerns like encryption, micro-segmentation, time-based retention methods like records management, and establishing a Recovery Point Objective.

If you look at your network configuration and blink an eye, it has changed. Listen to the interview to gather best practices from a data management specialist with decades of experience.

= = =

Want to leverage you next podcast appearance? https://content.leadquizzes.com/lp/fk1JL_FgeQ

Connect to John Gilroy on LinkedIn   https://www.linkedin.com/in/john-gilroy/

Want to listen to other episodes? www.Federaltechpodcast.com

Sometimes, the most important impact of technology can be behind the curtains.

Today, we will not focus on AI Deep Fakes and AI language translation, but on the impact AI on software development.

Terry Miller from Karsun begins the interview with the basics of software development: collecting user stories. Traditionally, a systems analyst would interview several users to see how a software system should behave.

This time-consuming process can be assisted with AI helping in assembling the details. Please note, Terry Miller will argue that AI is used as an auxiliary to humans putting together the stories. It can reduce time by eliminating tedious tasks and allowing well-paid professionals to focus on the key concepts.

Every large organization, including the federal government, has legacy systems that need to be replaced. This is always challenging because older systems have been put together with the proverbial duct tape. It can challenge a human to review code that has so many twists and turns to understand what to send to the new system and what is unnecessary.

AI can be used to review existing code to see where it can be leveraged to make an easy transition. Enterprise architects can then design a systematic approach to enable this transfer.

If you would like to see an example of this transition technology, you can visit Karsun Technologies and evaluate their tool called ReDuX-AI. 

= = =

Want to leverage you next podcast appearance? https://content.leadquizzes.com/lp/fk1JL_FgeQ

Connect to John Gilroy on LinkedIn   https://www.linkedin.com/in/john-gilroy/

Want to listen to other episodes? www.Federaltechpodcast.com

Ep. 150 Can a platform provide a path to safe AI for federal technology?

About every reader knows that ServiceNow is a large company that helps the federal government with managing large systems.

Today, Dr. Raj Iyer answers the question – is ServiceNow keeping up with AI to help reduce cost and improve delivery for massive federal systems?

Dr. Iyer is in a unique position because of his years of experience in commercial systems and a recent stint as the CIO of the United States Army. An argument can be made that he is up to date with managing large systems and has a thorough understanding of challenges in data security and cyber vulnerabilities.

During the interview, he provided a wide range of examples of how ServiceNow is leveraging its platform-as-a-service offering. A big splash was made when ChatGPT was able to generate content from large language models.

When it comes to the federal. government, they cannot use the stand large language model because of security concerns.

One approach that ServiceNow has taken is a partnership with NVidia that will all federal agencies to create domain-specific data sets. They can be compliant, secure, and allow innovation to be applied to a myriad of federal-specific problems.

= = =

Want to leverage you next podcast appearance? .

Connect to John Gilroy on LinkedIn   https://www.linkedin.com/in/john-gilroy/

Want to listen to other episodes? www.Federaltechpodcast.com

Everyone has seen AI with articles and even deep fake videos. No doubt it is dazzling, great for a conversation at a picnic with friends.

The question is . . . can you put AI in work boots?

Vishwas Lele has a novel idea: combine the best of AI with the best of proposal writing. Further, his innovation can be applied to an environment where data must be held securely. Yes, we are talking about proposals for the federal intelligence community.

Vishwas describes the Microsoft approach to AI – differentiates between free and paid Chat GPT as well as the free and paid versions of CoPilot.  He describes how Microsoft CoPilot can provide FedRAMP high protection for data.

When it comes to writing proposals for the federal government, we have several factors to consider:

1  AI depends on data libraries. If the proposal must include reviewing classified data, then precautions must be taken with forming data libraries.

2 Proposal writing is an art, not a science. By now, we have all heard of “hallucinations” provided by AI – my personal favorite is Napoleon being part of the American Civil War.

Vishwas suggests AI but must be amended with human knowledge. Shipley & Associates have garnered a stellar reputation in the proposal community.

 pWin.ai is partnering with Shipley and Associates to bring a human understanding of requirements to the proposal process.

Look for yourself. Go to pWin.ai and request a demo.

= = =

Want to leverage you next podcast appearance? .

Connect to John Gilroy on LinkedIn   https://www.linkedin.com/in/john-gilroy/

Want to listen to other episodes? www.Federaltechpodcast.com