ICE IT shop looks to build ‘agentic software factory’

Executive Summary

ICE has issued an RFI seeking an "agentic software factory" to use AI agents across much of the software development lifecycle, expand its existing STELLA platform, and require vendors to provide AI application development, workforce training, continuous monitoring, and technology evaluation services. Responses are due October 16, 2024. The RFI explicitly asks vendors to show how to differentiate genuine vendor capabilities from AI-generated proposals, signaling heightened scrutiny of claims about AI automation and provenance.

Market segments named in the Tags — IT Services; Artificial Intelligence; Software Development; DevSecOps; Training and Education; Homeland Security — are all directly implicated. Contractors in those segments should treat this RFI as a near-term business-development signal: prepare to show demonstrable technical capability, compliance-ready approaches (tags include NIST 800-171 (NIST Special Publication 800-171) and FedRAMP (Federal Risk and Authorization Management Program)), and clear methods for proofing authenticity of vendor deliverables. The event shifts procurement emphasis toward integrated AI-enabled toolchains, auditable pipelines, and workforce enablement tied to an existing agency platform (STELLA), so qualified bidders with integrated offerings that combine secure deployment, monitoring, and training can be advantaged.

Impact Matrix

IT Services

• Risk Level: High
• Opportunity: Support for AI-enabled application development, integration with ICE's STELLA expansion, and managed services for continuous monitoring and evaluation. Relevant NAICS codes: 541511, 541512, 541519, 541611, 541690, 611420. Agency/vehicle references: ICE / DHS (Department of Homeland Security); STELLA. Compliance surfaces to account for: NIST 800-171, FedRAMP.
• Timeline: Responses due October 16, 2024.
• Action Required: Assemble cross-functional proposals that combine systems integration, secure cloud or hosted services compatible with FedRAMP considerations, and demonstrable continuous monitoring capabilities; prepare artifacts showing how deliverables are produced and verified.
• Competitive Edge: Offer end-to-end integration plans that show auditable provenance, monitoring, and remediation workflows tied to STELLA integration and mapped to NIST 800-171/FedRAMP controls.

Artificial Intelligence

• Risk Level: High
• Opportunity: Provide agentic AI components, model development and evaluation frameworks, and methods to prove capability provenance and authenticity. Relevant NAICS codes: 541511, 541512, 541519, 541611, 541690, 611420. Agency/vehicle references: ICE / DHS; STELLA. Compliance surfaces: NIST 800-171, FedRAMP.
• Timeline: Responses due October 16, 2024.
• Action Required: Prepare technical demonstrations and documentation showing model training/testing, evaluation metrics, and mechanisms to differentiate genuine work from AI-generated claims; include explainability and validation steps.
• Competitive Edge: Emphasize auditable ML lifecycle controls (provenance, data lineage, versioning) and independent evaluation approaches that address ICE’s stated need to distinguish real vendor capability from synthetic submissions.

Software Development

• Risk Level: High
• Opportunity: Deliver developer tooling, agent-driven code generation/workflow automation, and integration into ICE’s software lifecycle for STELLA enhancements. Relevant NAICS codes: 541511, 541512, 541519, 541611, 541690, 611420. Agency/vehicle references: ICE / DHS; STELLA. Compliance surfaces: NIST 800-171, FedRAMP.
• Timeline: Responses due October 16, 2024.
• Action Required: Show secure development processes, CI/CD pipelines with audit trails, and ways to validate that produced code meets security and provenance requirements.
• Competitive Edge: Provide reproducible proofs-of-concept and developer workflow artifacts (logs, commits, signed builds) that demonstrate human oversight and traceability across the software lifecycle.

DevSecOps

• Risk Level: High
• Opportunity: Build secure, automated deployment and monitoring pipelines that support agentic automation and continuous monitoring requirements. Relevant NAICS codes: 541511, 541512, 541519, 541611, 541690, 611420. Agency/vehicle references: ICE / DHS; STELLA. Compliance surfaces: NIST 800-171, FedRAMP.
• Timeline: Responses due October 16, 2024.
• Action Required: Document how security is baked into agentic workflows, how monitoring/alerting will operate, and how provenance and attestation will be maintained across deployments.
• Competitive Edge: Present hardened DevSecOps patterns with demonstrable attestation, continuous monitoring, and automated compliance checks that map to the named compliance surfaces.

Training and Education

• Risk Level: Medium
• Opportunity: Provide workforce training programs and upskilling tied to operating and governing agentic software factories, and to supporting ICE personnel working with STELLA. Relevant NAICS codes: 611420 and others listed. Agency/vehicle references: ICE / DHS; STELLA. Compliance surfaces: NIST 800-171, FedRAMP.
• Timeline: Responses due October 16, 2024.
• Action Required: Prepare curricula and training delivery models for practitioners and administrators emphasizing secure operation, oversight of AI agents, and verification practices to distinguish authentic deliverables.
• Competitive Edge: Offer role-based, scenario-driven training packages that link directly to proposed technical solutions and audit processes.

Homeland Security

• Risk Level: Medium
• Opportunity: Support ICE mission needs through secure, auditable AI-enabled software development and operational capabilities; integration into the STELLA platform is central. Relevant NAICS codes: 541511, 541512, 541519, 541611, 541690, 611420. Agency/vehicle references: ICE / DHS; STELLA. Compliance surfaces: NIST 800-171, FedRAMP.
• Timeline: Responses due October 16, 2024.
• Action Required: Align proposals to mission context, emphasize security and authenticity controls, and be prepared to demonstrate how solutions meet operational constraints and sensitivity expectations.
• Competitive Edge: Combine domain-aware technical solutions with clear assurances (monitoring, evaluation, training) that address ICE’s specific concern about authenticity of vendor claims.

Cross-Segment Implications

• AI capabilities will drive requirements across Software Development, DevSecOps, and IT Services: agentic automation introduces new provenance and auditability needs that must be coordinated across development, deployment, and monitoring disciplines.
• Training and Education is a force-multiplier: effective upskilling enables sustained operation and governance of agentic systems and supports the auditable human oversight ICE is seeking.
• Homeland Security mission requirements and the STELLA platform act as the integration point: proposals will need to demonstrate both technical integration and compliance posture (NIST 800-171, FedRAMP) across segments.
• ICE’s explicit request for methods to differentiate genuine vendor capabilities from AI-generated proposals increases the value of independent evaluation, reproducible artifacts, and attestation mechanisms that cross all named segments.