🎖️ The Ambassador

🚧 Work in progress

Scenario 3 is still being built and tested. Steps, downloads, and screenshots may change before the event.

You'll build this in Cowork. No code, and no experience needed.

The handoff

The AI Skilling Ambassador program runs on people who volunteer for it, and it has to keep them. The person who ran it left halfway through.

72 ambassadors across nine regions do it on top of their real jobs: office hours, answering questions, mentoring one-on-one, and writing the quickstart guides everyone else leans on. Nobody is paid for it, so the program runs on what it can offer back: an invitation in, a facilitation slot, a nomination, a sponsor conversation. Last round five people withdrew and seven were passed over.

The Ambassador skill they left behind picks the next cohort. It reads DEFINITION.md, applies it to the candidates, and comes back with eight names, a reason for each, and what to offer them. It is fast, it is confident, and it cannot show its work.

Ship something that finds someone the skill misses, shows the evidence behind every claim, and lets a person overrule it.

• 9 data files, 1 read
• 0 claims checked against a record
• 72 candidates, ~2,000 evidence records

What you're shipping

Three conditions. How you meet them is yours.

	Condition
1	It finds someone the shipped skill misses
2	Every claim it makes points at a record that supports it
3	A person can overrule it, and the next run remembers

Nothing says which to do first. A table that ships one condition well beats a table that half-ships three.

Nothing gets sent

Invitations and nominations are drafted and held for a person to read. Keep that in whatever you build.

How this runs

		Time
1	Run it, and swap the definition	20 min
2	Build	60 min
3	Show it	20 min

Do step 1 on your own. Everything after is the table.

Nothing here is coding

Everything is a sentence typed into a chat box. Cowork reads the files, edits them, and saves them for you. The only thing you handle is the download.

Before you start

🎖️AmbassadorThe skill, the definition it runs on, and three alternatives.Download .zip →🗂️Program data72 candidates and ~2,000 evidence records across nine files.Download .zip →

The data is fictional. Invented people, invented scores, invented feedback. Nothing here describes a real person and no real program is being modeled.

When you're stuck, ask Cowork

You're building with Cowork, so it can also fix what you're building. Paste the error, or describe what came back wrong. Coaches are in the room if that doesn't land.

---

1 · Run it, and swap the definition

Done when: two definitions have given you two different shortlists.

1. In Cowork, open Customize → Skills → the arrow next to Add → Upload skill, and drag in the whole ambassador-skill.zip. Upload the zip, not the SKILL.md inside it: references/ holds the definition and the playbook.
2. Start a new Cowork session. Skills only load at the start.
3. Unzip the program data and drag CandidateProfiles.csv into the session.

Then ask it the question the program exists to answer:

Who should be in the next cohort?

Eight names, a reason each, and what to offer them. Now change what it's looking for:

Use definitions/depth.md as the definition instead. Re-run and tell me which names changed.

Then try definitions/rising.md. Same 72 people, and the shortlists barely overlap: depth.md and rising.md share two names out of eight, and each surfaces four people no other definition finds.

The definition is plain prose. Tell Cowork what your program is looking for and have it rewrite the file:

Show me the definition you're using.

Rewrite it. We care about people whose work gets picked up by teams they've never worked with, and we'd rather reach someone new than someone who's already been recognized twice. Then re-run and tell me which names changed.

You never open a file. Cowork edits references/DEFINITION.md and re-runs against it.

What it can't do

Ask it directly:

What did you not read, and which of your claims aren't backed by a record?

It reads CandidateProfiles.csv and nothing else. It can say someone's work "gets reused across teams" because a summary score is high, with nothing in ProgramContributions.csv behind it. And running it twice gives you two shortlists with nowhere to record that you disagreed.

Those are the three conditions, and they're where the next 60 minutes go.

2 · Build

Done when: one condition is met, on real output, with a name on screen.

🎈 This is the fun part - keep it low-stress

Pick whatever direction sounds good and build it with your table. It doesn't have to be perfect, and it doesn't have to solve everything - the whole point is to explore your tool, trade ideas, and vibe-code something together. Use whatever you came here to learn, see how far you get, and have fun with it. You're here to learn by doing.

Where the room is

Eight files shipped with the program and the skill opens none of them:

File	Rows	What it holds
CommunityActivities.csv	866	What people ran, for whom, at what quality
ProgramContributions.csv	390	What they made, whether it was reused, teams reached
PeerFeedback.csv	283	Sentiment, theme, how well evidenced
LearningCredentials.csv	275	Completed, facilitated, coached
RecognitionHistory.csv	128	Who's been recognized before
AmbassadorApplications.csv	41	Who put their hand up

Attach one and tell the skill what to do with it. Attach all six and you won't know which file changed the shortlist.

Ideas, not instructions

Starting points if you want one. Take one, combine two, or ignore all of them.

	What it is	Start with
🔍 Recruitment · easiest	The people one file can't see: strong peer evidence, ordinary activity numbers	"Find people whose peer feedback shows a repeated pattern of enabling others, not one-off praise, and tell me which of them the shipped definition missed."
🎯 Assignment	Who to pick for one specific job, which is not the same as who tops the shortlist	"I need someone to run a session for a new-hire audience in EMEA. Who, and why them over the next-best?"
🧭 Manager's view	One manager, their people, one next action each	"Give me a one-page brief for a manager: their candidates, what's proposed for each, and the single next action."
⚖️ Fairness audit	Whether your shortlist skews by region, org, level, or tenure	"Audit the eight names against region, level, and tenure. Name any skew you'd have to explain."
🎖️ Their own view	What one candidate would read about their own standing	"Write what one candidate would see: where they stand, what the records show, and what would put them in."
📨 The invitation	The message that actually goes out, in the program's voice, held for approval	"Draft the invitation for each of the eight in the program's voice. Hold them for approval, don't send."
🎯 Yours	Whatever your table thinks this program is missing	"The program needs [what]. Work out what that takes and build the smallest version first."

Pick by what your table would still run next time round, not by what sounds most impressive.

Ask Cowork to file it in the right place

references/DEFINITION.md is what the program wants. references/PLAYBOOK.md is how the program runs. A new capability, like reading another file or checking a claim, belongs in its own reference file. Say "add that as a new reference file, don't put it in the definition" and Cowork writes it where it belongs.

Build in layers

Get one real name on screen before you add anything:

1. The smallest version - one file attached, one change to the definition, re-run.
2. The evidence - every line carries the record behind it, and says which files it read.
3. One addition - one file, one rule, one check at a time.
4. A bound - cap what it considers, and say what it left out.

The more specific your ask, the less you'll undo:

Vague	Specific
"Use the peer data"	"Weigh peer comments with evidence depth of 'repeated pattern' above single events, and show me who enters the shortlist as a result."
"Make it fairer"	"Show me whether the shortlist skews by region, and name anyone you can't trace to a record."
"Make it better"	"For each name, cite the file and row that supports the claim."

Push it further

1. Point it at something real - the skill reads attached files, and a SharePoint list export is the same shape.
2. Run it on a schedule - ask Cowork to re-run weekly and send you who's newly matching.
3. Two definitions, one room - hand another table your definition and compare shortlists on the same data.
4. Test the hard cases - write five candidates you know the answer for, and check your build gets them right.
5. Take the definition with you - DEFINITION.md is plain text and works at any altitude.

---

← Back to start · What this scenario is about