Never Mind! HR Company Lattice Decided to Bring AIs Into the Org Chart, Then Decided Against It – Executive Summary
Inc., by Ben Sherry, published July 2024
1. Summary
“Never Mind! HR Company Lattice Decided to Bring AIs Into the Org Chart, Then Decided Against It” (Inc., 2024) documents how HR platform Lattice announced and then abandoned, within three days, a plan to give AI agents official employee records. Clients could have onboarded agents, set goals for them, and assigned them a direct manager, with Lattice's CEO arguing that treating agents like employees would let businesses use them while holding them accountable to company goals. The appeal was obvious: agents do not take breaks, speak many languages, can be in many places at once, and cannot unionize. What broke the plan was the reaction. Critics called the move tone-deaf and degrading to staff, one AI founder writing that it implies leadership sees people as resources to be measured against machines, and after hundreds of negative comments Lattice reversed course. One question was left standing, namely who is accountable when autonomous agents act.
2. Relevance for CIOs
Agents that hold system access, pursue goals, and make concessions on a company's behalf are already running in most enterprises, usually without a named owner. Persistence levels, discount matrices, and concession boundaries are permission and control design, so they belong with identity, access, and audit. The org chart is a side debate. What every deployed agent needs is a named human owner, a defined scope, a lifecycle, and a cost line – either within the enterprise architecture or as a teammate.
3. Thoughts from Sonny Claussen (CEO Bee360) about this Article
Sonny sees the Lattice reversal as the right finding paired with the wrong lesson, and he points out that treating AI like a person tends to make humans less accountable. His view is that the mistake was putting AI into the org chart without governance underneath it. The test he uses in client conversations comes down to a single question: does the AI just execute, or does it act on its own? If it executes, he treats it as technology that belongs in the enterprise architecture with an application owner, a lifecycle, and a cost line, which is territory every IT organization already knows how to govern; if it acts, he treats it as a teammate that needs what any new hire needs, meaning a named human owner, a job description written as guardrails, clear priorities, and a budget, with token consumption serving as its salary.