Luke EatonData Driven Recruitment

Issue 73 3 min read

Airbnb's AI for HR Playbook

Howdy recruiters! Check out Airbnbs new playbook!

Howdy Recruiters!

I'm writing this while chasing princesses around Disneyland with my daughters so excuse the brevity.

Mary Kate Stimmler attended a talk last week from Iain Roberts at Airbnb, and posted the best summary I've seen of what an actual, working AI-for-HR operating model looks like.

Five concrete moves Airbnb is making right now.

This is the playbook. Let's go!

Trade your PDFs for .md files

Markdown. Plain text. Machine-readable.

Every policy doc, every interview guide, every playbook, every job spec template. Out of the locked PDF, into a markdown file that an LLM can actually parse, reason over, and quote back to you accurately.

You cannot have an AI-enabled TA function on a foundation of locked PDFs.

This is the cheapest, highest-leverage move on the list. Do it this week.

Turn your best presentations into "skills"

Take the decks your TA function has built over the years. Your best onboarding deck. Your hiring manager training. Your interview calibration session.

Convert them into a structured "skill" that an AI agent can call.

The agent doesn't need to relearn how Airbnb does interview calibration every time it gets asked. The skill is the canonical reference. Build it once, reuse it forever.

For recruiters this means: your hiring manager training scales infinitely overnight

Use meeting recordings to measure rhetoric-to-reality distance

This one's spicy.

Record your meetings. Transcribe them. Use AI to compare what leaders SAY they want, with what they actually approve, hire, and promote.

Distance = your culture problem.

For TA specifically, run it on intake calls. Compare what the hiring manager said they wanted on Monday with what they actually shortlisted on Friday. The gap tells you everything.

Leaders must not delegate. They must build.

This is the one I've been banging on about for months.

If you're a TA leader and you've delegated the entire "AI strategy" thing to a junior on your team, you've already lost.

You learn by building. You learn the limits, the prompts, the workflows, the failure modes. Then you can lead.

Stop reading thinkpieces. Pick a workflow this week. Build it yourself.

Hire and train "embedded organisation architects"

This is a new role.

An organisation architect sits inside a business function and redesigns the workflow for an AI-native world. They're part operator, part engineer, part change manager.

In TA, this is the person who looks at your sourcing → screen → interview → offer flow and rebuilds it with AI in the middle. Not someone who slaps a chatbot on top of your existing process.

If you're a recruitment leader, you should be writing this job description right now. Or arguing that the role exists in your team already and just needs a new title.

Pick one.

If you can only do one thing this week, do number 1. Convert one policy doc to markdown. Stick it in a folder your team can access. Tell people to use that one instead of the PDF.

its not sexy, but it compounds.

Luke Update

Out of office this week. I'm at Disneyland with the family, eating overpriced churros and queuing for things.

Replies will be slow. If the world ends, please leave a note.

See you next Wednesday.

Luke

Well... that's issue seventy three! If you have any questions about it, or any feedback on this issue of The Data Driven Recruiter, grab me on LinkedIn for a chat.

I'll see you next week!

All 77 issues →