Salesforce has expanded Missionforce, its AI platform for government and national security organizations, with new capabilities and partnerships with OpenAI and NVIDIA designed to help agencies deploy mission-specific artificial intelligence while maintaining control over sensitive data and infrastructure.
The expansion gives government agencies additional tools for customizing AI, automating policy execution, coordinating field operations and modernizing administrative processes. Salesforce is positioning Missionforce as an AI environment that can operate across secure government infrastructure, including private cloud and air-gapped networks.
Through Salesforce Government Cloud, OpenAI’s frontier models will integrate with Salesforce Public Sector Solutions through Amazon Bedrock. Missionforce applications and workflows are also planned for ChatGPT, allowing authorized government and national security personnel to work with approved data and trigger Missionforce actions through a conversational interface.
Salesforce said the combination will enable agencies to reason across mission data and execute multi-step workflows while keeping access and actions within controlled government environments.
The company is simultaneously expanding its work with NVIDIA. Salesforce plans to bring NVIDIA models and accelerated computing into Missionforce so organizations in highly sensitive sectors can train, tune and deploy mission-specific models using their own critical data.
Salesforce and NVIDIA are also working to bring post-trained NVIDIA models into Missionforce and allow organizations to run specialized models on their own infrastructure.
A major component of the expansion is the new Missionforce Policy Engine, which uses generative AI to turn approved policy documents into structured rules that can be incorporated into government workflows.
OpenAI models will be used to generate deterministic rule code from approved policy documents and create test cases to validate whether those rules behave as intended. Salesforce said all generated outputs will remain subject to human review and approval before deployment.
The Policy Engine is also expected to integrate with ChatGPT, allowing citizens to ask policy-related questions while helping government employees retrieve current guidance more quickly.
Salesforce illustrated the capability with a state Medicaid scenario. If a family’s income changes, the system could automatically reevaluate a child’s eligibility under current rules and generate an audit trail showing how the determination was reached.
Another new capability, Missionforce Operations, is designed to convert manual administrative processes such as procurement, supplier management and invoice auditing into digital workflows.
Specialized AI agents can coordinate tasks, monitor status and identify exceptions in real time. These workflows can run within private cloud or air-gapped environments, an important requirement for defense and national security organizations that cannot depend on conventional public-cloud connectivity.
Salesforce plans to fine-tune NVIDIA models specifically for Missionforce Operations. These models can be trained around an agency’s internal terminology and operational data so AI agents can reason through administrative processes while remaining inside the organization’s infrastructure.
In one example, Salesforce said Missionforce could support a military or government fleet during a maintenance surge by scanning inventory across multiple depots, locating available spare parts and creating transfer manifests and dispatch schedules. The company said such automation could potentially reduce equipment downtime from days to minutes.
Salesforce is also extending Missionforce into field operations and asset management.
Missionforce Field Operations & Asset Management uses AI agents to automate scheduling, work orders and maintenance while supporting native offline mobile capabilities for teams operating in locations without reliable connectivity.
Following a natural disaster, for example, AI agents could automatically generate follow-up work orders for inspectors and emergency response teams, helping agencies coordinate assessments and repairs involving electricity, water and other public infrastructure.
The broader strategy reflects Salesforce’s effort to adapt agentic AI to environments where security, auditability and operational control are particularly important.
Rather than relying entirely on one model or infrastructure provider, Missionforce is being designed to combine Salesforce’s government applications and workflow technology with third-party AI models and computing infrastructure.
OpenAI provides frontier model capabilities, while NVIDIA contributes models and accelerated computing technology that can be customized and deployed closer to sensitive government data.
The approach also allows agencies to keep humans involved in consequential decisions. The Policy Engine, for example, can use AI to interpret and structure approved government policies, but Salesforce said the resulting rules and tests must still be reviewed and approved before they are deployed.
Missionforce’s expansion comes as government organizations increasingly evaluate generative AI for citizen services, administrative automation, defense operations and decision support while also confronting requirements around security, data sovereignty, regulatory compliance and explainability.
Salesforce is positioning Missionforce as the layer connecting those AI models with the workflows, data and operational systems agencies already use.
KEY QUOTES:
“Government agencies want tailored AI that runs everywhere they operate while staying within the secure environments their missions demand. Missionforce enables this at scale, with purpose-built AI that executes the mission and enables agencies to solve their hardest problems while maintaining operational control.”
Kendall Collins, CEO of Missionforce & Government Cloud
