HR / Training OS
Stop rebuilding every onboarding plan, policy answer and programme from scratch.
Your week is built from other people’s requests. The onboarding plan for Monday’s new joiner, the manager who wants training for the team, the fourth question this month about the flexible-work policy. Each one gets produced by hand, from scratch, because the knowledge that makes it good lives in your memory and in files scattered across drives, inboxes and system admin panels.
This course builds you a people and learning brain in the Claude Code desktop app: a governed memory that holds how your organization develops people, all traceable to sources. Claude prepares, checks and surfaces what needs you, through a weekly People Pulse and skills you build once and reuse. Every decision about a person, and every send, stays human, on the record, and yours.
No terminal, no code, no tech skills. You work in a simple desktop app, in plain language, and every change waits for your approval before anything happens. On Claude Pro your work is private and is not used for training. Your employee data stays in your HR systems; the OS works from approved documents and sanitized snapshots, and it never stores an individual record.
Built on the Claude Code app. Tailored to your organization. No coding. No tech skills. For beginners.
7 sessions · €149 + Claude Pro subscription (€18/month, required)
The hiring manager wants the role profile by Friday. A manager wants communication training for the team. A new joiner is waiting on their onboarding plan. The same policy question arrives for the fourth time. Every request gets produced by hand, from scratch, and the quality depends on what you happen to remember that day.
Most AI-for-HR training makes this worse in a specific way: it speeds up producing documents without a memory underneath them. The output gets faster, the function does not improve, and every artifact starts from zero again.
The teams pulling ahead run a different setup. Their function has a memory: how the organization develops people, with every claim traced to the source that approved it. Claude prepares onboarding, answers policy questions from the actual clause, helps managers get ready, and diagnoses a training request before designing anything. The people decisions stay human.
You build that in the Claude Code desktop app, around your real roles, policies and programmes, without being technical and without writing a line of code. And the two real risks of AI in HR, employee data leaking and fluent output quietly making people decisions, are exactly what the course is built to prevent.
- Your roles and the capabilities they carry
- Your onboarding standards
- Your policies, explained from the source
- Your training and learning programmes
- Your managers and the conversations they run
- Your HR and learning data, sanitized and aggregate
- Your privacy boundaries, written down first
- Your way of building capability, not just delivering courses
By the end, you have a working system that prepares onboarding to a standard, answers policy questions from the exact clause, gets managers ready for the conversations they own, diagnoses a training request before designing anything, and gets sharper every time you use it.
That is how you move from reacting to the people work to designing it.
No AI experience needed. No coding, no tech skills. Made for beginners.
Three layers: your approved sources, a wiki Claude writes and keeps current, and the rules that govern how your function prepares, checks and escalates. You read it. Claude maintains it.
- Set up the Claude Code desktop app and stand up your workspace in the first session, guided click by click
- A raw sources layer of approved policies, role descriptions and programme materials, read only, as your source of truth
- A compiled wiki where every claim traces to the file that approved it, and new formulations are marked until you approve them
- Twelve operating rules (your CLAUDE.md), including the ones that keep individual employee records out and every people decision human
No terminal, ever. You work in the desktop app and describe what you want in plain language.
Role expectations become observable behaviours a manager and an employee can both point to, and the checks that keep them fair run on everything.
- Role profiles where every expectation is an observable behaviour: what standard, strong and falling short look like on the job
- An evidence check that separates fact from interpretation from recommendation, and names what a decision-maker still needs
- A fairness check that flags assumed or inferred protected characteristics and proposes a neutral rewrite
- A capability map that ties your business priority to the capabilities and roles that carry it
Only your approval turns a proposal into an organizational standard. The checks then run inside every later skill.
The brain starts reading where the function lives, and the week gets a heartbeat that surfaces only what needs you.
- Scoped, read-only connectors for a people-requests label, one approved-documents folder and your calendar, each with its own governance card
- A sanitized export-drop pattern for HR and learning data: the personal columns come off at the source, the snapshot answers one question, then it is deleted
- A weekly People Pulse that shows exceptions, not a dashboard, and names patterns, never individuals
- A memory gate that decides, with you, what the week teaches the wiki and what stays temporary
Your HR systems stay the record. The OS reads only what you scope, and it never sends anything.
Define what good looks like once, with the failure modes and escalation points written down, then let the work run to it.
- An onboarding-designer that plans day 1, week 1, month 1 and month 3 with an owner for every item, and flags anything contractual to Legal
- A policy-explainer that answers from the exact clause and says so where the policy is silent, never determining compliance
- A manager-conversation-prep that gets the manager ready and never assesses the employee or drafts a verdict
- An employee-communication that drafts plainly with every claim cited, and where a named human always sends
You define good before you automate, so you never automate inconsistent people work faster. You approve every output, and nothing is ever sent.
From delivering courses to building capability with evidence, diagnosed before it is designed.
- A training-needs-analysis that classifies a request and asks whether training is even the right response before anything is built
- A programme-designer that designs along the capability chain, from need to objective to practice to reinforcement to evidence
- A learning-evaluation that defines success before delivery and reports in aggregate, never grading an individual
- One programme designed end to end, with its evaluation plan written before anyone is trained
The diagnosis, not the course catalogue, decides what gets built. Capability is measured as a change you can see.
One programme, taken end to end, then the loop that makes every future one start sharper.
- One full cycle from need to decision to evaluation, with the pack assembled from the skills you built
- A decision record that answers, a year later, where a requirement came from and who decided
- A programme-retrospective that proposes what the wiki, the standards, the skills and the rules should permanently learn
- The accept-or-reject pass where you decide, item by item, what the organization keeps
The cycle is finished when the function has learned from how it landed, and the learning is governed by the same rules as everything else.
By the end of session seven, you have:
- 1 people and learning brain in Claude Code
- 13 reusable Claude Skills
- 1 weekly People Pulse
- Role profiles and a capability map for your key roles
- 1 designed learning programme with its evaluation plan
- 1 complete cycle with a decision record
And you walk away with:
- A people function whose knowledge compounds with every programme
- A way of working where you decide and Claude prepares and checks
- The knowledge to keep expanding your OS on your own
Move from reacting to the people work to designing it.
7 sessions · €149 + Claude Pro subscription (€18/month, required)
For HR managers, people teams, learning and development professionals, trainers, organizational development teams