Alvys launches AI agents for freight TMS workflows
Alvys launched Foundry, an agentic AI platform that embeds autonomous AI agents directly within its transportation management system (TMS) for carriers and brokers to automate freight workflows The platform offers over 20 pre-built agent templates covering detention processing, document handling, rate audits, shipment tracking, asset compliance, and claims management, plus custom agent creation via plain-language descriptions or uploaded SOPs Foundry includes a governance layer called Agent Shie
Analysis
TL;DR
- Alvys launched Foundry, an agentic AI platform that embeds autonomous AI agents directly within its transportation management system (TMS) for carriers and brokers to automate freight workflows
- The platform offers over 20 pre-built agent templates covering detention processing, document handling, rate audits, shipment tracking, asset compliance, and claims management, plus custom agent creation via plain-language descriptions or uploaded SOPs
- Foundry includes a governance layer called Agent Shield with approval thresholds, spending limits, human sign-off for high-impact actions, and a full audit trail, alongside a model-selection system that routes tasks across LLMs based on cost, speed, and quality
- The launch reflects a broader industry shift where TMS platforms are evolving from systems of record into proactive execution platforms, as seen with similar agent deployments at C.H. Robinson and Uber Freight
- Alvys raised $77 million in total funding (including a $40 million Series B in September 2025) and processes over $9 billion in freight annually, with Foundry rolling out through controlled customer cohorts
Why It Matters
This represents a significant step in the adoption of agentic AI within enterprise logistics, moving beyond AI that merely analyzes or generates content toward systems that autonomously execute operational tasks within existing workflows. For AI practitioners and logistics professionals, it demonstrates how embedding agents directly into domain-specific platforms—rather than layering on separate tools—can reduce friction, eliminate redundant logins, and leverage rich contextual data already present in the system.
Technical Details
- Agent Architecture: Foundry supports both pre-built templates (20+) and custom agents that operators can build by uploading standard operating procedures or describing tasks in plain language, with the platform generating workflows for approval
- Pre-Built Agents: Include Detention Agent (detects exceeded detention time and files claims), Document Intelligence (reads and files rate confirmations, bills of lading, proofs of delivery), Track & Trace Agent (handles check calls and shipment updates), Rate Audit Agent, Asset Compliance Agent, and Claims Agent
- Governance & Security: Agent Shield provides configurable approval and spending thresholds, human sign-off pauses for higher-impact actions, and a complete audit trail of agent decisions and manual overrides; the platform is SOC 2-compliant with agreements preventing customer data from being used to train public models
- Model Routing: A built-in model-selection system dynamically routes tasks between large language models based on cost, speed, and quality criteria, avoiding vendor lock-in and managing computing expenses
- Integration Context: Foundry operates on Alvys' existing TMS infrastructure with 120+ integrations and native EDI connections to hundreds of shippers, giving agents access to lane history, customer rules, documents, margins, appointments, and exceptions
- Testing & Deployment: Agents can be tested against simulated data before live deployment, with operators able to build, deploy, monitor, and pause agents through the Foundry platform
Industry Insight
- TMS platforms are becoming execution engines: The industry is moving rapidly from passive systems of record to proactive platforms that automate operational work—Alvys, C.H. Robinson, and Uber Freight are all pursuing this trajectory, signaling that future TMS competitiveness will depend on agentic capabilities
- Embedded agents outperform standalone AI tools: By keeping agents within the existing TMS rather than introducing separate tools and logins, Alvys addresses a common failure mode in enterprise AI adoption where new tools disrupt rather than integrate with established workflows
- Governance is a differentiator: Features like Agent Shield—with configurable thresholds, human-in-the-loop controls, and audit trails—will be critical for enterprise adoption in regulated industries, and companies that bake governance into their agent platforms will gain a competitive edge over those treating it as an afterthought
Disclaimer: The above content is generated by AI and is for reference only.