Tuesday, ten past two
Mark runs an electrical wholesaler out of Geelong. Sixty people, three branches, a trade counter that opens at 6.30. In August his ops manager set up an AI assistant on the shared drive so the counter crew could ask it questions instead of ringing head office.
The first real test came on a Tuesday afternoon. Josh, three weeks into the job, had a sparky at the counter wanting $4,000 of cable on an account already $1,800 over its limit. Josh asked the assistant what to do.
It gave him three answers. The credit policy from 2021 said stop supply and refer to accounts. An email from the branch manager last summer said regulars could run 30 days over. And the values page from the last offsite said "We back our trade customers."
Josh did what any sensible new starter does. He rang Shaz in accounts, who has been there nineteen years and knows exactly who pays. She said yes. The sparky got his cable.
Nothing went wrong that day. But Mark now had a printout showing three versions of his business, all written by his own people, none of them agreeing. The assistant wasn't broken. It read exactly what the leadership team had left behind.
The warning signs
That's what dead culture looks like when AI joins the team. Not a scandal. A quiet gap between what's written down and what really happens, held together by people like Shaz.
In most businesses we see, it shows up in five places.
1. The values are on the wall, but they never come up in a decision. Think about when someone last said "we're doing this because of our value about X" in a real meeting. If nobody can remember, the values are decoration.
2. Nobody can say the plan back without notes. Ask five people what the three priorities are this year. If you get five different lists, or a lot of looking at the ceiling, there isn't a plan. There's a document.
3. The real rules live in one or two heads. Every business has a Shaz. When the answer to "how does this work?" is a person's name, the rule isn't written anywhere.
4. Priorities get added, never dropped. The 2024 project is still technically live. So is the one from the offsite before that. Half the business is still working on things leadership quietly stopped caring about.
5. The same question gets different answers. Ask two branches, two shifts or two team leaders how a common situation is handled. If the answers differ, your customers have already noticed.
None of these signs mean people are doing the wrong thing. Usually they're doing the sensible thing with what they've got. The friction is in what leaders never wrote down, or wrote down three different ways.
The gap
Culture is what people do at two o'clock on a busy Tuesday. It isn't what's on the wall.
Every business has three versions of itself. What leadership says. What's written in the policies, procedures and systems. And what actually happens on the floor. When those three line up, new people learn fast, customers get the same answer twice, and you can take a week off.
When they drift apart, the business still runs. It just runs on goodwill and memory. People like Shaz become the system. That works right up until Shaz takes long service leave.
Here's how far apart the versions can get. In one 80 person distribution business, we asked 12 leaders and team leaders to write the three company priorities on a sticky note. We got 19 different priorities, and only two people wrote the same three. Demo Every one of them had been in the room when the priorities were set.
The AI test
This is where AI changes things. Not because it's clever, but because it's literal.
A new person works out the unwritten rules in a few months. They watch, they ask, they learn which policy is real and which one is old. An AI tool can't do that. It reads what it can access: your decks, your policies, your shared drive, your inbox if you let it. Then it acts on that, quickly, without the judgement to say "that one's out of date".
So it shows you the gap. Ask it for your priorities and see what comes back. Ask it how you handle a late-paying customer. Ask it what your values mean for a refund. Whatever it says is a fair reflection of what you've actually written down.
That's useful. AI doesn't create dead culture. It shows it to you, faster. And as AI tools start acting on their own, approving credit, answering customers, booking jobs, the gap stops being something you can see and starts being something that happens to a customer.
How we'll look at the whole business
Over the next eleven weeks we'll go through the business one part at a time, roughly in the order the problem travels: leadership, customer service, team leaders, sales, marketing, operations, finance, people, technology and delivery. In week 12 we put it all on one page, your culture map.
Each week asks the same five questions. You can run them on any team today.
- What does the company say? The values, priorities, policies and promises.
- What's actually written down? The procedures, systems and decision rules people and AI tools can find.
- What really happens? The shortcuts, exceptions and "ring Shaz" moments.
- What happens when AI gets involved? Where a literal reader would get it wrong.
- Where does it travel next? Which team, customer or system catches the problem.
The connection
This week isn't about one department, but the pattern always starts somewhere and moves.
In Mark's case it started in the leadership team, which wrote a values line and never squared it with the credit policy. It passed through finance, whose policy was right on paper and ignored in practice. It landed on customer service at the counter, where a new starter had to pick between three answers. And it keeps going into sales, because the reps have been telling builders "we'll look after you" on accounts finance wants stopped.
Keep that chain in mind. Most of the friction in this series shows up downstream and starts upstream.
The ident check
Answer yes or no. Nobody's marking it.
- Could five people from different parts of the business name the same three priorities without notes?
- Could each of them say what their own part is in those priorities?
- Could they say what the business stopped doing this year?
- Has a company value been used out loud to make a real decision in the past month?
- If your most experienced person left tomorrow, would the rules they carry be written down anywhere?
- Would two sites, branches or shifts handle the same common situation the same way?
- If you asked an AI tool on your shared drive for your priorities, would it give one clear answer?
Questions 1 to 3 are the recall test. If any of those is a no, start there. Everything else makes more sense once you know what people actually carry around in their heads.
Cheap first
- Ask three people the recall questions this week. Costs nothing, takes a day. Pick someone senior, a team leader and someone new. Write down exactly what they say.
- Find your Shaz. List the two or three people whose name is the answer to "how does this work?" Don't change anything yet. Just know who they are.
- Run the AI test. If you have an AI assistant on your drive, ask it for your three priorities and one common policy question. Print the answers.
- Count your live priorities. Every project, initiative and focus area still technically running. If the number is above five, you've found part of the problem.
Then the harder fix: getting the leadership team to agree on three, write them down in words a team leader would use, and announce what's been dropped. That's next week.
The Monday close
This week, ask three people to name your three priorities without notes. Don't correct them. Just listen and write it down.
Next week we start where it usually starts: the leadership team. Priorities nobody remembers, decisions that never leave the room, and what happens when an AI tool starts reading the plans.
Questions people ask
- How do you know if your company culture only exists on paper?
Look for five signs: values that never come up in a decision, a plan nobody can say back without notes, rules that live in one or two heads, priorities that get added but never dropped, and the same question getting different answers in different teams.
- Does AI cause culture problems?
No. AI tools read what you've written down and act on it literally. They don't create the gap between what you say and what you do. They show it to you, faster.
- What is the recall test?
Ask people from different parts of the business, without notes: what are our three priorities, what's your part in them, and what did we stop doing this year? If the answers differ, the plan hasn't reached them.
Read the episode transcript
The Ident Series, Episode 1: What dead culture looks like when AI joins the team
Show notes
Episode title: What dead culture looks like when AI joins the team
Summary: Mark's electrical wholesaler switches on an AI assistant, and a new starter gets three answers to one credit question. Trav uses the story to set up the 12 week series: the five signs a culture only lives on paper, why AI tools show the gap faster, and the five questions every episode asks.
- Three versions of one business: what leadership says, what's written down, and what really happens
- The five warning signs, from values that never come up in a decision to the same question getting different answers
- Why AI is literal, not clever, and why that makes it useful
- The ident check: seven yes or no questions, including the three question recall test
- This week's job: ask three people for your priorities, without notes
Full article: getsorted.valuebull.com.au/ident/dead-culture-meets-ai/
Script
G'day. It's Trav, and this is The Ident Series. Twelve weeks, one part of the business at a time, working out where your culture actually lives. And what happens when AI tools start reading it.
This is episode one. What it looks like when the culture's only on paper, and AI joins the team.
I want to start with a story. It's made up, but I reckon you'll recognise it.
(pause)
Mark runs an electrical wholesaler down in Geelong. Sixty people, three branches, trade counter opens at half six. Back in August his ops manager set up an AI assistant on the shared drive. The idea was simple. The counter crew could ask it questions instead of ringing head office all day.
First real test was a Tuesday afternoon. Josh, three weeks in, has a sparky at the counter wanting four thousand dollars of cable. Problem is, the account's already eighteen hundred over its limit. So Josh asks the assistant what to do.
And it gives him three answers.
The credit policy from twenty twenty-one says stop supply, refer to accounts. An email from the branch manager last summer says regulars can run thirty days over. And the values page from the last offsite says, we back our trade customers.
So Josh does what any sensible new starter does. He rings Shaz in accounts. Shaz has been there nineteen years. She knows exactly who pays. She says yes. Sparky gets his cable. Everyone goes home.
(pause)
Nothing went wrong that day. But think about what Mark's holding now. A printout with three versions of his business on it. All written by his own people. None of them agreeing.
The assistant wasn't broken. It read exactly what the leadership team had left behind.
(pause)
That's what dead culture looks like when AI joins the team. It's not a scandal. It's a quiet gap between what's written down and what actually happens. And it's held together by people like Shaz.
Here's how I'd put it, if you were sitting next to me in the car. Culture is what people do at two o'clock on a busy Tuesday. It isn't what's on the wall.
In most businesses I walk into, the gap shows up in five places. Let me run through them, and you see which ones sound familiar.
One. The values are on the wall, but they never come up in a decision. When did someone last say, we're doing this because of our value about such and such, in a real meeting? If you can't remember, they're decoration.
Two. Nobody can say the plan back without notes. Ask five people for the three priorities this year. If you get five different lists, or a lot of looking at the ceiling, there's no plan. There's a document.
Three. The real rules live in one or two heads. Every business has a Shaz. When the answer to how does this work is a person's name, the rule isn't written down anywhere.
Four. Priorities get added and never dropped. The project from twenty twenty-four is still technically live. So's the one from the offsite before that.
Five. The same question gets different answers. Two branches, two shifts, two team leaders. If they'd handle the same situation differently, your customers already know.
(pause)
Now, none of that means your people are doing the wrong thing. Mostly they're doing the sensible thing with what they've got. The friction is in what leaders never wrote down. Or wrote down three different ways.
So why does AI matter here?
Not because it's clever. Because it's literal.
A new person works out the unwritten rules in a few months. They watch, they ask, they figure out which policy is real and which one's old. An AI tool can't do that. It reads what it can get to. Your decks, your policies, your shared drive. Then it acts on that, quickly, and it doesn't know which bits are out of date.
So it shows you the gap. That's actually useful. AI doesn't create the problem. It shows it to you, faster. And once these tools start acting on their own, approving credit, answering customers, booking jobs, the gap stops being something you can see. It becomes something that happens to a customer.
(pause)
Over the next eleven weeks we'll go through the business piece by piece. Leadership, customer service, team leaders, sales, marketing, operations, finance, people, technology, delivery. Then in week twelve we put it on one page.
And every week, the same five questions. What does the company say. What's actually written down. What really happens. What happens when AI gets involved. And where does the problem travel next.
In Mark's case, it started with leadership, who wrote a values line and never squared it with the credit policy. It went through finance. It landed on a new starter at the counter. And it's heading into sales, because the reps have been telling builders, we'll look after you.
Shows up downstream. Starts upstream. Remember that one.
(pause)
Right. Here's this week's ident check. Seven questions, yes or no. I'll go slow. Nobody's marking it.
One. Could five people from different parts of your business name the same three priorities, without notes? (pause)
Two. Could each of them say what their own part is, in those priorities? (pause)
Three. Could they say what the business stopped doing this year? (pause)
Four. Has a company value been used out loud, to make a real decision, in the last month? (pause)
Five. If your most experienced person left tomorrow, would the rules they carry be written down anywhere? (pause)
Six. Would two of your sites or shifts handle the same common situation the same way? (pause)
Seven. If you asked an AI tool on your shared drive for your priorities, would it give you one clear answer? (pause)
The first three are the recall test. If any of those is a no, that's where I'd start.
So here's what I'd do this week. And it costs nothing.
Pick three people. Someone senior, a team leader, and someone new. Ask them to name your three priorities without notes. Don't correct them. Don't explain. Just listen, and write down exactly what they say.
If you want a second job, find your Shaz. Work out who the two or three people are whose name is the answer to, how does this work. Don't change anything yet. Just know who they are.
(pause)
Next week, we go to where it usually starts. The leadership team. Priorities nobody remembers, decisions that never leave the room, and what happens when an AI tool starts reading your plans.
I'm Trav. This has been The Ident Series. See you next week.