Share with your CHRO
Hitachi Digital bet that HR fragmentation, not headcount, was its real productivity problem. Spanning five distinct companies, 20-plus systems of record, and wildly inconsistent policies across business units, the organization was averaging five-plus days to resolve a single HR query. The company deployed an AI HR companion called Skye, built by Ema, that reasons across fragmented policy documents and takes direct action inside ServiceNow, Jira, and Okta. Early projections show 50-70% time savings on HR activities and a 30% month-over-month drop in ticket volume.
What this means for your business
The Hitachi case is most instructive for CHROs whose HR complexity comes from M&A history or multi-entity structures rather than raw headcount growth. If your org has consolidated companies without consolidating systems, you’re carrying the same fragmentation tax Hitachi diagnosed: queries that require cross-platform context take days because no human can hold all the policy variations in their head simultaneously. The question isn’t whether AI could help, it’s whether your fragmentation is visible enough to scope a deployment around it.
The deployment detail that actually matters here is the distinction Ema CEO Surojit Chatterjee draws between a chatbot and an agent. A chatbot retrieves; an agent reasons and acts. Skye doesn’t just surface a leave policy, it files the request. That action-layer is what collapses resolution time from days to hours, and it requires deep integration work, not a prompt wrapper on top of a knowledge base. The article notes that Hitachi explicitly rejected the “a couple of engineers can stitch this together” assumption, which is the most expensive mistake HR tech buyers make in agentic AI right now.
The cultural packaging matters more than most technical post-mortems admit. Giving the agent a name, a personality, and an explicit “co-worker” framing isn’t soft strategy; it’s the mechanism that drives adoption before the tool has had enough reps to prove itself on accuracy alone. Skye earned employee trust through consistent performance, but the design choices bought the time needed to accumulate those reps. CHROs who skip that framing step and deploy a nameless, brandless bot are optimizing for procurement optics while sabotaging the adoption curve that makes the ROI real. I’d revise this view if adoption data from generic-branded deployments showed comparable engagement, but nothing in the current market suggests that.
Based on reporting from AI in HR done right: enterprise lessons from Hitachi, originally published 2025-07-15 03:00:00.
