Somewhere between 60% and 80% of federal AI projects have stalled out and the common denominator almost always traces back to the data. In Episode 16 of The GIST of Govt IT, Brian and Sean sit down with Matt Lawson, Director of Solution Engineering for NetApp Federal, to unpack what it actually takes to build a data foundation that lets AI succeed. Matt's thesis is that projects keep stalling because teams can't get the right data, in the right place, at the right cost.
The conversation digs into data gravity (you can't move petabytes overnight, and you can't beat the speed of light), the take-the-data-to-the-model vs. bring-the-model-to-the-data debate (answer: you need both), and the copy-of-a-copy sprawl that quietly wrecks both budgets and data authority. Matt walks through ideas that change the math, and the guys close out on post-quantum cryptography and why “harvest now, decrypt later” makes it a today problem, not a tomorrow problem. Plus Matt on why he always chooses the blue pill.
Featured Guest
• Matt Lawson, Director of Solution Engineering, NetApp Federal
The “Why AI Projects Fail” Data
AI, RAG & Partners
• Retrieval-Augmented Generation (RAG) explained
• Apache Kafka (real-time event streaming)
Post-Quantum Cryptography
• NIST Post-Quantum Cryptography Standards (FIPS 203/204/205)
• OMB M-23-02 (migrating to post-quantum cryptography)
The Monday-Morning Playbook (Matt Lawson)
• 1. Catalog your data — know what data sets you actually have
• 2. Plan for a federated environment — data will live in many places, some not even yours
• 3. Design for security from day one — immutability, ransomware recovery, and PQC can't be bolted on later
Related Episodes
• Episode 9: Quad Charts Be Damned: Data Meets the Mission
Upcoming Event
• GIST 360 Breakfast Briefing with NetApp — Washington, DC, November 10
The Hosts & Show
• Swish