Jeremy Wilcox

What’s Jeremy been up to? That’s probably what you are thinking, or I owe you a text or email response. How do I enter the public square? Let’s try a softball topic. If you know me, I don’t deal with softballs! Fastball incoming, high and tight. The next fight in government AI is not about who has the best agent or the best model. It is about who owns the control layer — private companies, or the American people. That may sound like architecture-speak. It isn't. When AI agents start acting across government systems, the control layer is what decides what they can see, what they can do, what rules they follow, which models they use, and what gets recorded when they act. In the agentic era, that's where the actual authority lives. The ontology becomes the dictionary. The model layer becomes the brain. The runtime becomes the workplace. The audit trail becomes the record of accountability. If all of that lives inside a private vendor's platform, the government still looks like it's in charge on paper. In practice, it isn't. On November 30th, 2022, I was working for USDS out of the White House, detailed to DHS, and tried out GPT 3.5 after work. I simulated a master calendar hearing with EOIR in front of a judge, and the fact it had absorbed the Immigration and Naturalization Act was impressive. It felt like April 22nd, 1993 times ten. It got a number of things wrong, but I caught those mistakes. We are coming up on another inflection point. The American people do not elect SaaS companies. They do not elect foundation model providers. They do not elect cloud platforms. They elect representatives who write the laws, fund the agencies, oversee the execution, and answer for what gets done with public authority. So as the federal government moves into agentic AI, I think one line has to be non-negotiable. The control layer belongs to the public. Vendors can help build it. They should not own it. That means the government owns the mission logic. The government owns the model gates. The government owns the audit trail. The government owns the authority to decide when AI agents act, stop, escalate, or explain themselves. This is not an anti-vendor argument. It is a pro-accountability argument, a pro-Constitution argument, and a pro-public-trust argument. Vendors who build the right architecture will thrive in this environment. Vendors who try to own what should belong to the public will eventually find themselves on the wrong side of a congressional hearing, a GAO opinion, or a federal court ruling — and the procurement officers who are paying attention already know this. Like for Part 2, as I share how Microsoft's collaboration software was better in 1996 when I worked there, than it was at the White House in 2023, and why that's relevant now, in 2026.

If there's one thing we should all be able to agree on... May the 4th be with you!

89% growth in Federal bookings! Let's gooooo!

This is HUGE at HHS. When my son was paralyzed with a rare 1 : 2,000,000 condition in 2018, I was told there was a systemic breakdown of information sharing and analysis at every level (including state and local). No one had answers. It was one of the reasons I answered the call to federal public service at USDS in 2019. When I saw C3's Agentic AI Platform for the first time, I immediately saw the potential to solve some of the toughest challenges in government. Glad to be a part of the C3.ai team!

It's been great hearing from friends, former colleagues, and extended family members who returned to work today in the federal government -- or have been working the entire time, and will finally be getting paid. I appreciate you, and thank you for your service! When it comes to AI, if you're like me, you want to see real-world examples of how government agencies are using AI today. On December 2, we're hosting a live webinar with Carahsoft to do just that. The conversation will cover tangible use cases that improve public safety, simplify operations, and deliver measurable value to agency users and their constituents, without complex setup or steep learning curves. AI doesn’t need to be mysterious or hard to implement — it’s already driving results for agencies of all sizes. Register here: http://spr.ly/60487k9Rw

On Friday, I was honored to attend the 2025 Service to the Citizen Awards event in Washington, DC. It was amazing to celebrate so many great accomplishments with so many great people! The DHS CX Directorate was well represented as a Category Winner! It was inspiring to stand alongside so many public servants and innovators who are raising the bar for how government serves people. For me, it was also a moment of personal reflection, taking me back professionally to 2011 when Martha Dorris was my customer (thanks for the unexpected shoutout!) Martha was the first public servant in the federal government to join the Forrester Leadership Board for CX professionals. I remember walking into Martha’s office and being blown away by the CX research, toolkits and strategies in action at GSA! Our meeting with GSA Administrator Dan Tangherlini helped cement the work you had been pioneering. Although I stepped away from the CX space for a time, I never stopped following the progress. Friday’s event was proof of just how far this movement has come. Congratulations to all of this year’s awardees! Thank you for showing what’s possible when we keep citizens at the center of government service delivery! Meet the 2025 Winners: https://lnkd.in/eUrth-9p STTC Website: https://lnkd.in/eWRRhRrb P.S. I’m still available to work next month. Reach out if you are looking to level up your team’s impact heading into FY26! Nicshan Floyd Kelly Olson Dominic Sale Casey Coleman Dustin Brown Alexandra (Ali) Tobolsky MaryAnn Monroe, CCXP Rishi Vajpeyi   #STTC #ServiceToTheCitizen #GovIT #DigitalServices #CustomerExperience #Innovation #PublicService #CX #SBIR