# People Team AI

> Building AI capability in People Teams. We take a real problem, solve it with AI,
> and publish the whole thing: the working tool, the prompts that made it, and what
> it cost to run. Free. Every build ships as a public repo with a
> README.md and a PROMPTS.md.

Roughly six of every ten pieces are tutorials, two are reusable skills, one is a peer
interview, and one is a tool review or benchmark score. This is not a review site;
reviews are the smallest part of the output.

Position: some things are worth buying and some are a day's work. We do not argue
against buying software. We teach People Teams to tell the difference, and publish the
small builds that do not need a subscription.

Every entry is tagged with what you need before you can follow it:
- setup: none. A chat window you already have. Nothing is installed.
- setup: 10-min. A saved project with instructions and files. Still no install.
- setup: one-off. An editor and a coding assistant, set up once and reused.
- setup: hosted. Somewhere to run it, and somewhere to remember things if needed.
Most of what a People team needs is setup: none, which is the point.

Entries are tagged by level so a reader can find work that fits where they are:
- beginner: concepts and getting set up; small tasks with quick wins.
- intermediate: can set things up, add a database, host a simple app.
- advanced: full multi-user builds at the scale that replaces paid software.
Reference material such as the benchmark carries no level.

## Research
Sourced writing on the numbers, methods and rules behind AI at work. Every claim carries
its source, its year and its sample size. Where a widely quoted figure turns out to be
unsupportable, that is stated rather than quietly omitted.

- [The HR Assistant question](https://peopleteamai.com/research/hr-assistant-exposure/) [4 sources]: Human Resources Assistants have the second largest "needs software built first" band of any People occupation, at 68 per cent against a Job Zone 3 average of 30. Only 11 per cent of the role is reachable with a chat window today, which means the advice being sold to these people, learn prompting, is aimed at two tasks out of nineteen. It is also one of only two People occupations you can enter without a degree, and an independent Tufts model puts it at the highest projected job loss of the nine. The risk is not a redundancy round. It is a role that quietly stops being hired, which removes the way in rather than the people already through it.
- [The 70% statistic](https://peopleteamai.com/research/seventy-percent-transformation-failure/) [3 sources]: The 70% figure traces back to Hammer and Champy in 1993, who described it as an unscientific estimate and later disowned the fixed-rate reading. A peer-reviewed study in 2011 followed the number across five published sources and found no empirical basis in any of them. McKinsey states it without a source, the Harvard Business Review version traces to a magazine column, and BCG's own breakdown shows 30% full success and 44% partial value, which is not a failure rate. If you need a defensible number, McKinsey's 2018 finding that 16% of digital transformations both improved performance and sustained it is the better one.
- [The 95% AI pilot claim](https://peopleteamai.com/research/mit-95-percent-ai-pilots/) [4 sources]: MIT Project NANDA's State of AI in Business 2025 found that around 5% of custom enterprise AI tools reached production with sustained productivity or P&L impact. That is not the same claim as 95% of GenAI pilots failing, which is how it was reported. The evidence base is 52 executive interviews, 153 survey responses collected at four conferences, and a review of 300 or so publicly disclosed initiatives. The report is not peer reviewed and its method has been publicly challenged. The finding worth your attention is a different one: externally sourced tools reached deployment around 67% of the time against 33% for tools built in house.
- [Measuring an AI pilot](https://peopleteamai.com/research/measure-ai-pilot-worked/) [3 sources]: A pilot that started without a baseline cannot prove anything, because there is nothing to subtract from. UK government benefits management doctrine, published in the Teal Book, requires that current performance is captured before work starts, that every benefit has a named owner, and that monetisable benefits are separated into cash-releasing and non-cash. Applied to an AI pilot, that means hours released and cash converted are different figures that must never be added together, and anything a vendor claims stays in a third column until a baseline makes it measurable.

## Live sessions
Free online webinars and interviews. Built live, questions answered as we go, recordings
published free afterwards with the repo and the prompts.

- Nothing scheduled yet.

## Published
- Nothing published yet.

## Planned
- [Automation risk report](https://peopleteamai.com/library/automation-risk-report/) (advanced) [setup: hosted]: A task-level exposure report built from a job description. It parses the JD into tasks, maps them to O*NET task statements, applies a published versioned rubric, and returns which tasks are exposed, which are durable, and where to invest in training. It never identifies who to make redundant.
- [Advert checker](https://peopleteamai.com/library/job-advert-checker/) (beginner) [setup: none]: Paste a job advert and get back the specific lines that narrow your applicant pool, with the reason for each. Not a rewrite tool. It shows the cause so you can decide what to change.
- [HR tool benchmark](https://peopleteamai.com/library/hr-tool-benchmark/) [setup: none]: A published method for scoring HR and talent tools. Every score we publish cites the benchmark version it was produced under, and the rubric lives in a public repo so anyone can check the method or disagree with it in the open.
- [Exit interview synthesiser](https://peopleteamai.com/library/exit-interview-synthesiser/) (beginner) [setup: none]: A skill that reads a folder of exit interview notes and returns themes with counts and supporting quotes. The point is the count. A theme mentioned once is an anecdote, and a theme mentioned nine times is a problem.
- [Write your own skill](https://peopleteamai.com/library/write-your-own-skill/) (intermediate) [setup: 10-min]: The guide rather than the artefact. What a skill is, what makes one reliable, how to scope it to a single job, and the signs that what you actually need is a script or a tool instead of a skill.
- [Claude Code commands](https://peopleteamai.com/library/claude-code-cheatsheet/) (beginner) [setup: one-off]: A reference table of Claude Code's slash commands and keyboard shortcuts with a plain explanation of what each one does and when you would reach for it. Written for People teams rather than engineers, so it says what a command is for rather than what it technically invokes.

## Contact
- Contact: everything goes through https://peopleteamai.com/ask/. No email address is
  published, because a published address is a scraped address. Tool access for review
  never involves payment or commercial
  terms in either direction.
- Sponsorship: handled separately and never affects a score.