A $1.6B contract that reframes what agentic buying looks like
On July 24 the Department of Veterans Affairs awarded Salesforce a $1.6 billion Agentic Enterprise License Agreement, one of the largest public commitments yet to AI agents doing real operational work. The structure matters as much as the number. It is a one-year base with two optional one-year renewals, so the headline figure is a ceiling the VA has to earn its way toward, not a signed check. For CIOs watching from the private sector, that shape is the interesting part. It signals that even a marquee buyer wanted an exit ramp before betting three years on agents.
The scope is genuinely large. The agreement covers 170 VA medical centers and more than 1,100 outpatient clinics, serving over 17 million veterans, families, and caregivers. In 2025 the VA handled 82 million direct-care appointments, and VA Health Connect fielded 40.6 million veteran calls. Any agentic system that touches those volumes is operating at a scale most enterprises never reach. That is precisely why the deal is worth studying: it is a stress test of agentic software at a scale that will expose whatever breaks.
Missionforce sets the compliance floor for regulated agents
The contract runs on Missionforce, Salesforce's government solution, and the technical framing is a lesson in what regulated agentic work requires. Missionforce is FedRAMP High-authorized and its AI agents are described as HIPAA-ready. For any leader in healthcare, financial services, or the public sector, those two labels are the gate. An agent that can act on protected health data has to carry the same authorizations as the systems it touches, and Salesforce is leaning on that posture as its differentiator against every startup pitching cheaper agents without the paperwork.
This is where the build-versus-buy calculus gets sharp. A regulated enterprise could assemble its own agent stack, but replicating FedRAMP High and HIPAA readiness is a multi-year compliance program before a single agent runs. Salesforce is effectively selling the authorization as much as the automation. For CIOs, the question is whether that compliance shortcut justifies the platform lock-in, because once agents, data, and workflows all live inside one vendor's authorized boundary, moving off it becomes a governance project of its own.
The 28-days-to-minutes claim is the whole thesis
The operational goal is blunt: cut average appointment scheduling from 28 days to minutes. That single metric carries the business case. Veterans wait, staff wrestle with disconnected systems, and the agentic pitch is that automation collapses the coordination overhead that produces the delay. Kendall Collins, CEO of Missionforce and Government Cloud at Salesforce, put the logic plainly: "Every minute a VA employee spends navigating disconnected systems is a minute not spent serving a Veteran." The framing is efficiency recovered from integration failure, which is a story most enterprises recognize in their own operations.
We would caution readers to treat 28-days-to-minutes as a target, not a delivered result. Scheduling delays at the VA are rooted in capacity, staffing, and clinical constraints as much as software friction, and no agent conjures appointment slots that do not exist. The honest read is that agents can compress the coordination and navigation time, which is real, while the harder bottlenecks remain physical. CIOs should hold vendor productivity claims to that distinction and ask which part of a delay is genuinely a software problem.
Three fronts, one bundled license
The deal attacks three fronts at once: 24/7 virtual care coordination through AI agents, connected team workflows delivered in Slack, and unified data through MuleSoft and Tableau. That combination is the real product. Salesforce is selling agents bundled with the integration layer that feeds them and the collaboration surface where humans stay in the loop. The MuleSoft and Tableau inclusion is a tell: agents are only as good as the data plumbing behind them, and disconnected systems were the stated problem in the first place.
Dave Rey, President of Global Public Sector at Salesforce, framed the ambition around the institution's mission: "Missionforce gives them the tools to match that commitment at scale." For enterprise buyers, the packaging is the strategic signal. Bundling data unification, workflow, and agents into a single agentic license is how Salesforce raises switching costs and captures budget that might otherwise fragment across point vendors. It is an elegant commercial design, and one that every competing platform will now try to copy.
Consumption and renewal risk sit with the buyer
The renewal structure deserves a second look because it defines who carries the risk. A one-year base with two optional extensions puts the burden of proof on Salesforce inside the first twelve months. If the agents do not measurably move scheduling times and call handling, the VA can decline to renew and the $1.6 billion ceiling never materializes. That is a healthier procurement posture than the multi-year lock-ins that defined the last enterprise software era, and private-sector CIOs should borrow the pattern in their own agentic contracts.
The open question is cost predictability. Agentic pricing tends toward consumption, and at VA volumes, 82 million appointments and 40.6 million calls, consumption can scale in ways a fixed license never did. The AELA framing suggests Salesforce and the VA negotiated a ceiling to contain that, but the mechanism inside the license will determine whether costs stay legible. Any leader signing an agentic agreement should insist on the same clarity: know what an agent action costs before the volume arrives.
What this means for your agentic roadmap
This deal becomes a reference point whether or not it succeeds. When a federal agency with 17 million constituents commits up to $1.6 billion to agents, it validates the category for boards and gives every vendor a case study to cite. CIOs will be asked why their own agent programs are still stuck in pilots when the VA is deploying at national scale. The honest answer is that the VA bought authorization, integration, and a renewal-gated commitment, not magic, and that combination is reproducible only by matching the discipline behind it.
Our advice is to read the structure, not the headline. The instructive parts are the compliance floor Missionforce sets, the bundling of data and workflow with the agents, and the one-year escape hatch that keeps the buyer in control. Copy those into your own build-versus-buy decision. If you cannot articulate the compliance posture, the data plumbing, and the exit terms of your agentic bet with the same clarity, you are not ready to sign one at any price.
