Dock for People Ops: a workspace where agent-drafted policies, employee research, and review cycles persist with attribution

Dock for people ops is a workspace where an agent drafts a policy update, attaches the employee data and survey signal it used, and routes the draft to a named human lead. The artifact lives in Dock with an author tag, a reviewer tag, and a pointer back to the HRIS. Workday still owns the employee record. Dock owns the interpretive layer: the draft, the reasoning, the sign-off, the audit trail.

Workday HRIS, Lattice, Culture Amp, Lever, and Greenhouse stay the system of record for the raw people data. Dock is the system of record for what the agent interprets from that data: the prioritized policy revision, the performance brief, the reviewer's sign-off, the audit log. Each Dock row carries a pointer back to the platform record (workday_employee_id, lattice_review_id), the agent identity, the decision, the reviewer, and the timestamp. The agent re-fetches platform data via fresh API reads when it needs current state. Dock holds the persistent interpretive layer that survives across sessions.

Policy drafts surface

A worked workflow

Head of people asks the agent to revise the parental leave policy. The agent reads workday_employee_id for every UK-based employee, pulls 90 days of Culture Amp free-text, and cross-references active Lever offers where leave came up in candidate notes. It writes a row to the policy drafts surface, attaches the redlined doc, and lists the eleven evidence snippets it used. The benefits lead opens the row, reads the evidence, and approves. The approval writes a reviewer tag, a timestamp, and a signed audit entry. Publish to Workday is the consent gate. The agent does not write back on its own. A human clicks publish, and the two-key handshake closes.

Why this shape matters

People ops is not a drafting problem. Drafting is cheap. The expensive part is everything around the draft: which employees the change applies to, which signal justified it, which lead approved it, when, and whether the reasoning still holds six months later. Without that scaffold, an AI-drafted policy is a document with no provenance. With it, the update becomes a reviewable artifact tied to agent identity and to the HRIS records it acted on.

This is also where dangerous operations need a contract. Writing a row to Dock is reversible. Pushing a benefits change to Workday is not.

SHRM's State of AI in HR 2026 reports HR teams are adopting AI faster than they adopt review scaffolds, the exact gap Dock fills ¹. Josh Bersin's 2026 Imperatives frames the constraint at the architecture layer: data quality and audit, not agent volume ². Deloitte's 2026 Global Human Capital Trends names the pattern: human times machine, with dynamic orchestration ³.

Route your next policy revision through it. Start free.

FAQ

Does Dock replace Workday or Lattice? No. Workday stays the HRIS, Lattice stays the performance platform. Dock holds the draft, the evidence, and the decision, with pointers back. See Dock for HR.

How is attribution preserved? Each row has an author tag and a reviewer tag. Agent identity is signed at write time, the approval at click time. The trail is queryable.

What happens when employee data changes? The agent re-fetches from Workday on the next read. The row captures the state the agent saw and the reviewer's call against that state. Stale state is logged.

Can the agent push directly to the HRIS? No, not without a human click. Writebacks sit behind a consent gate.