Gallatin AI, Inc.

Speed of innovation. Speed of integration. Quality and availability of data. The DoW has set expectations for how they expect the industry to meet their needs. These same priorities shape how we build at Gallatin. We recently shared insights into our process and products when our CTO Daniel Buchmueller sat down with Chad Wahlquist to unpack how we're building on Palantir Foundry to improve defense sustainment outcomes.

Contested logistics will define the future of large-scale conflict. Autonomous resupply is one of the clearest opportunities to change the math on who bears that risk. That's what our recent work with Kodiak was built to demonstrate. Gallatin's Navigator detected a resupply need, generated an optimized convoy plan in under 90 seconds, and handed off directly to Kodiak's autonomous driving system that executed the mission. No soldier exposure on the route. Operators maintained full approval authority with live video streamed to handheld tactical devices. What remains is the work of scale: building the integration architecture that connects AI-driven planning to autonomous execution, through a common, secure data interface, across every echelon and service that needs it. That's where Gallatin is building. Full story below:

Contested logistics will define the future of large-scale conflict. Autonomous resupply is one of the clearest opportunities to change the math on who bears that risk. That's what our recent work with Kodiak was built to demonstrate. Gallatin's Navigator detected a resupply need, generated an optimized convoy plan in under 90 seconds, and handed off directly to Kodiak's autonomous driving system that executed the mission. No soldier exposure on the route. Operators maintained full approval authority with live video streamed to handheld tactical devices. What remains is the work of scale: building the integration architecture that connects AI-driven planning to autonomous execution, through a common, secure data interface, across every echelon and service that needs it. That's where Gallatin is building. Full story below:

Gallatin took second place at NVIDIA's Live Agent Gauntlet at GTC. The format: build and optimize an AI agent in advance, then load it live on stage and compete head-to-head on standardized benchmarks with results tracked on a real-time public leaderboard. Our CTO and co-founder Daniel Buchmueller signed up two days before the event, built the agent, and competed. Won round one. Close second in round two. Won round three. Advanced to the finals. Second place overall. At Gallatin, we build AI systems that deliver under constraints: compressed timelines, hard benchmarks, and problems that don't wait for perfect conditions. That's true whether it's an agent competition at GTC or generating optimized sustainment plans in under 90 seconds for a Brigade Combat Team. Short clocks. Hard benchmarks. That's where we operate. Looking forward to competing in the next one with NemoClaw!

I couldn't say it better than Isabelle Bousquette: One of my highlights here at #GTC was spending time in the shady Build-A-Claw tent, talking about and further experimenting with #OpenClaw (built by 🦄 Peter Steinberger) and #NemoClaw (built by Alexander Watson and team). I've been using #ClawdBot since January and my use case Isabelle mentioned in her article is just a sliver. For me, the three most fundamental aspects of this revolution are: 1. Async Nature of Computing Complementing the current synchronous chat agent interaction paradigm with an asynchronous one: Since claws run on machines that are always on, they can do slow and long research and inference work overnight and report back to me (as a cron job or when completed) in a repeatable and reliable fashion. 2. Universal Computer Use The evolution of computing has brought us to today where the majority of interactions with systems happen through input from keyboard, mouse on the desktop and touch on phones and screens as outputs. The easiest way for a computer program to interact with another system is through an #API, where the technical contract between the two systems is clearly defined and this is mostly how we've been operating. Recently, with the addition of #MCPs, capabilities or also called tools can be discovered more easily by #LLMs but fundamentally, they are still strict technical contracts. Now contrast this with how you use your computer every day: You can book a flight by navigating to your airline's website in your browser, enter origin and destination, compare prices and pay via credit card. You wonder how many parking spots are free in which garage in #SanJose so that you don't waste your time navigating to the wrong garage when you get to #GTC: You can log into the official website of garages here in San Jose and look for the garage with the most number of free spaces. For this, you actually have to click around through all garages and look at each garage's occupancy. #OpenClaw can use your computer (almost fully) as if it was you: It can send text and mouse inputs to websites and programs and read what's displayed on the screen and process and understand that. 3. Local Compute When using LLMs you traditionally talk to a large model hosted in the cloud and you pay per token consumed. With OpenClaw, you can take advantage of local compute to do that with smaller but still very powerful models for everyday tasks. In addition, you can also make use of local computing power to do deeper analysis (i.e. count cars in a picture using #YOLO). At Gallatin AI, Inc., we deeply believe that #NemoClaw can help transform not just enterprise but also warfighter applications in a fundamental way through careful introduction: A lot of defense logistics data isn't available through APIs but rather operated through "swivel chair APIs", aka by humans and their keyboards and mice. PS: Yes, I texted my claw to visualize the availability of garage spaces in downtown San Jose :-).

Sustainment planners in a contested environment don't just need better data. They need to predict what's needed and generate executable options faster than the situation changes. Gallatin AI has been awarded a Direct to Phase II SBIR contract under the Army Applications Laboratory's PORTAL program, bringing predictive logistics decision support to tactical sustainment. When routes are denied, assets are redirected, and the mission shifts, planners don't have hours to rebuild a plan. Navigator turns that crunch into seconds of executable options.

-30°F breaks your planning factors. Here's what we learned running Navigator in the Arctic for the first time. This winter, Gallatin deployed Navigator with the 17th CSSB, 11th Airborne Division at JPMRC 26-02 in Alaska. Vehicles idle continuously. Consumption spikes in ways that don't map to any temperate baseline. Standard planning factors stop working. We ran our predictive algorithms against real Arctic conditions for the first time and fed that data directly back into the platform. The 17th was already working across Maven, WICKR, Teams, and manual trackers. Navigator ingested their LOGSTATs and gave the commander a real-time supply picture every time levels changed — not just when reports came in. We committed to a data-driven AAR within 12 hours of ENDEX. Consumption variance, reporting latency, backhaul risk — built from what Navigator captured throughout the rotation, not reconstructed from memory afterward. Every rotation makes the models sharper. That's the cycle we're building. #ArmyLogistics #DefenseTech #Sustainment #ArcticOperations #MilitaryAI Full write-up here: https://lnkd.in/g9SkB5vB

The idea for Gallatin was born in Austin. It's officially home to our third office. Texas has more military installations than any other state. Fort Hood, Joint Base San Antonio, and Fort Bliss are all within reach. Transformation and Training Command (T2COM),  the Army's center of gravity for modernization, is headquartered here. The density of Army logistics activity in this part of the country is hard to match anywhere else. We build software for military logisticians. That work doesn't get better in a vacuum. It gets better through proximity to the commands, planners, and operators working these problems every day. Austin puts us inside that ecosystem, not adjacent to it. The Austin office will serve as one of Gallatin's engineering and delivery hubs. If you want to build technology that operational sustainment planners will actually use, we're hiring. More on why we chose Austin and what we're building there: https://lnkd.in/gHWiE8eX

Gallatin AI is hiring to modernize defense logistics with an AI-first approach: from factory to foxhole. The user base of our logistics AI decision-support system, Navigator, is growing rapidly, so we're putting out the call for an Engagement Manager to join our Delivery team. You'll own the customer relationship end-to-end — onboarding, delivery, expansion — and lead cross-functional teams to deliver AI/ML solutions to defense customers while driving real value for the U.S. military. If you're passionate about AI, logistics, and national security, it's time to join the fight: https://lnkd.in/gBmJ5ch9

At Gallatin, we're rebuilding how supplies move from factory to foxhole. We're hiring for three AI Engineering roles: Routing & Network Optimization – Build routing solvers for dynamic, capacity-constrained networks. When you optimize a route, it affects real resupply missions. https://lnkd.in/gKxUGK9D Decision & Optimization Systems – Own the constraint layer that enforces authority and policy. Your feasibility checks determine what plans are allowed to exist. https://lnkd.in/ga7qEEcB Allocation & Packing Systems – Solve vehicle loading and resource allocation at scale. Bridge the gap between AI-generated plans and physical reality. https://lnkd.in/gh9RiwP5 If you want your work to matter, let's talk.