Editorial illustration for Workday Builds Enterprise-Scale AI Agent Platform with LangGraph and LangSmith
Workday Launches Enterprise AI Agent Platform
Four engineers built the infrastructure behind Workday's agent push this year. That's the entire team tasked with standing up a platform to build, deploy, and run AI agents across a company that serves more than half the Fortune 500 in HR, finance, and IT. Workday's leadership set a target in 2025: agents in production by the end of 2026. The catch is that Workday runs under some of the strictest security and data governance rules in enterprise software, so every service has to clear audit, compliance, and retention checks before it ships.
The team's answer was an internal developer platform built on LangGraph for agent orchestration and LangSmith for tracing, observability, and evals. The scope was large for four people: a durable and scalable agent server, data access controls wired into Workday's audit policies, a collaborative eval system, and an SDK to wrap it all so product teams wouldn't need to touch the plumbing. The first product out of that effort was Workday's Agent System of Record, a governance layer that decides which APIs agents can actually reach. What that build process looked like, and why the small team structure mattered, is where the story gets specific.
Previously, building a prototype agent took 4 to 5 months, with no guarantee it could ever reach production. With the new platform, a team can have a prototype running in a staging environment in under a day, with authentication and security requirements baked in. The only question left for the agent-building team is: what does this agent actually do, and what does the graph look like?
Why this matters
A four-person team building the agent infrastructure for a company touching more than half the Fortune 500 tells you something about where agent tooling actually stands in 2025. Workday didn't write its own orchestration layer from scratch. It bet on LangGraph and LangSmith and built an abstraction layer on top, handling auth and deployment so internal teams could focus on what their agents actually do. For founders weighing build-versus-buy on agent infrastructure, that's the real signal: the orchestration and observability layer is becoming commodity enough that even a company Workday's size didn't see value in owning it end to end.
The harder question is what happens when this platform has to serve hundreds of agent teams instead of a handful. Lean teams can ship fast, but Workday now has two constituencies, its own engineers and the customers running those agents in production, and both will expect reliability most startups never have to deliver. Worth watching whether LangChain's tools hold up as the dependency graph gets more complicated, and whether a team of four scales headcount to match.
Related Reading
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
From Archives Google Launches Gemini Decisions API at USD 0.10/M Tokens, 50-100 ms Compute
From Archives US corporate AI spending hits a new high in October, growth slows to 1.65%
Common Questions Answered
How did Workday reduce prototype development time for AI agents using LangGraph and LangSmith?
Workday reduced prototype development time from 4-5 months to under a day by building an abstraction layer on top of LangGraph and LangSmith that handles authentication and security requirements automatically. This allows internal teams to focus on defining what the agent does and designing the graph structure rather than spending months on infrastructure and compliance work. The new platform bakes in security and audit requirements from the start, eliminating the uncertainty of whether a prototype could ever reach production.
What was Workday's production timeline goal for deploying AI agents across the enterprise?
Workday's leadership set a target to have AI agents in production by the end of 2026. This ambitious goal was established in 2025 and represents a significant commitment to integrating agent technology across their enterprise platform that serves more than half the Fortune 500 companies.
Why did Workday choose to build on LangGraph and LangSmith instead of developing its own orchestration layer?
Workday's small four-person infrastructure team recognized that building a custom orchestration layer from scratch would be inefficient given the company's strict security and data governance requirements. By leveraging LangGraph and LangSmith as foundational tools and building an abstraction layer on top, the team could focus on solving Workday-specific challenges like authentication and deployment while relying on proven agent tooling infrastructure.
What security and compliance constraints does Workday's AI agent platform need to address?
Workday operates under some of the strictest security and data governance rules in enterprise software, meaning every service must clear audit and compliance requirements before deployment. The new agent platform addresses this by baking authentication and security requirements directly into the infrastructure, ensuring that all agents built on the platform automatically meet these stringent enterprise standards without requiring additional compliance work from individual teams.
How many engineers were responsible for building Workday's enterprise-scale AI agent platform infrastructure?
Only four engineers built the entire infrastructure behind Workday's AI agent platform push, which is tasked with enabling agent development, deployment, and operation across a company serving more than half the Fortune 500. This small team's ability to create a production-ready platform demonstrates the maturity and effectiveness of modern agent tooling like LangGraph and LangSmith in 2025.
Further Reading
- How Workday Built an Enterprise-Scale AI Agent Platform with LangGraph and LangSmith - LangChain
- Introducing Workday's Trusted Agent Platform - Workday
- Workday cuts AI agent prototyping from months to less than a day - AI News Bites
- How Lyft Built a Self-Serve AI Agent Platform for Customer Support with LangGraph and LangSmith - LangChain
- The Best AI Agent Frameworks in 2026 - LangChain
