Beta · free while we learn

Synthetic HR data,
one POST away.

The API humans and agents call when they need a Workday-shaped demo dataset — employees, events, comp, recruiting — without touching real people's data.

$ claude mcp add --transport http \

peoplesets https://www.peoplesets.com/mcp

Approve in the browser. No key copy-paste, no env vars. Works in Claude Code, Cursor, Cline.

Get a free key →Read the docs

or see the actual rows it produces ↓

Built by Steven Shoemaker — People Analytics at Deel.

0

industry packs

0

scenarios

0

countries

0/60

fidelity checks

Industry packs

Pre-shaped narratives

tech_startup, retail_chain, healthcare_system — each pack is a curated config with the right department weights, attrition rates, and geo mix baked in.

Scenarios

Layered storylines

rif, hyper_growth, m_and_a, distressed, leadership_shake_up. Drop one onto any base and the events table shows it cleanly.

Deterministic

Reproducible by contract

Same call, same dataset. Forever. Demos that don't drift between meetings, pipelines that don't break on re-run.

The promise

Same call,
same dataset.
Forever.

Most demo data is a one-off. Peoplesets isn't. You can re-run the exact request next week and get byte-identical output: every employee, every termination, every salary change. The same way the same Python script gives you the same answer twice.

The same request always returns the same dataset.
Every manager ID resolves. Every salary fits its pay band.
New fields get added. Existing fields never change shape — your downstream code keeps working.
Employees live in real countries with real cities and the right currency.

Built for agents

A real Claude session, end to end.

Install the skill, ask in plain English. Claude picks the pack, calls the API, polls the job, hands you the artifact. No prompt engineering required.

Not a Claude user? The same endpoints work from curl, Python, or anything that speaks HTTP.

claude session — peoplesets skill installed

youretail chain, ~5k employees, US and Canada, frontline-heavy.

↳ toolPOST /generate-company → industry_pack: retail_chain, country_mix: {USA: 0.85, CAN: 0.15}, seed: 42

↳ toolGET /jobs/719b3a41ce64 → status: done · 4,458 active · 11,191 terminations · 26% frontline turnover/yr

claudeBuilt a 4,458-person retail chain. Frontline (L1/L2 Ops + CSM + Sales) shows realistic 26% annualized churn. Artifact: /tmp/719b3a41ce64.zip

See the output

Real data, not a mockup.

Pick an industry, layer a scenario, see actual rows from a real engine run. Download the full 500-employee CSV — no key needed.

Parameters

Industry

Scenario

Fixed for the preview: starting headcount 500, 3-year window, seed 42. The final CSV includes every hire and termination across the window — typically 2–6k rows depending on the pack’s turnover and growth profile.

The full API exposes 40+ knobs — size up to 10k, country mix, multiple scenarios, custom seeds.

Get started

Free. No card. 30 seconds.

Drop your email — we mint a key inline and show you the first call. Use it from the Claude skill, curl, or any HTTP client.