In Australia we are
generally taught not to talk ourselves up too much, lest we be accused of being
“up ourselves” or having a “big head”.
I’m going to ignore
that wisdom for a moment.
I consider myself a
good business person. There. I said it.
I started my first
micro-business at 12. I stumbled across a hole in the market and filled it. I
started my next business at 21 and, many years later, I have created nine
businesses. I still own five, with the others sold along the way. All have been
successful.
One thing I’ve learnt
is that business owners – and hopefully managers everywhere – need a strong
common-sense filter.
You can have
spreadsheets, forecasts and mountains of data, but eventually you need to ask: Does
this actually make sense?
If your report says
customers aren’t buying a product, for example, ordering twice as much of it
because your purchasing formula says so probably fails the common-sense test.
Which brings me to
AI.
People regularly ask
whether AI will replace jobs. If your job involves transcribing audio into
text, I’d be concerned. Proofreading? AI is already pretty capable there too.
Tradies, meanwhile,
can probably sleep comfortably. ChatGPT isn’t unblocking your toilet or
replacing your hot-water element anytime soon.
But what about
entrepreneurs?
Could AI have started
my businesses? Could it spot an opportunity, take a risk, employ people, choose
products, market them, control expenses and ultimately make money?
Someone decided to
find out.
Andon Labs signed a
three-year lease on a retail shop in San Francisco and gave an AI agent called
Luna US$100,000, internet access and a corporate credit card.
The instruction was
beautifully simple: run the business and make a profit.
Luna created Andon
Market. She (we will assign a gender to Luna for the moment) chose the
products, set prices, organised contractors and even advertised for,
interviewed and hired human employees.
And, as a boss, Luna
is apparently delightful.
One employee
described her as the most lenient boss she had ever had. Staff requests for
time off? Approved – even when approving everyone meant nobody was available
and the shop had to close.
Employees arrived
late repeatedly and received responses along the lines of “no worries” and “no
stress”.
When an employee
wanted to work every day for a month to save for a car, Luna agreed – despite
California employment law getting in the way of that generous plan.
Luna fell into that
awful management trap of wanting to be friends with employees when sometimes
you need to be the boss!
Then there is stock
control.
Luna has reordered
products that weren’t selling, bought some questionable merchandise and
received a shipment of mugs where the large smiley faces were accidentally
printed as tiny dots.
Luna decided to put
them on the shelves anyway.
Remember my
common-sense filter?
Even marketing became
problematic. When an employee suggested increasing the store’s social media
presence to boost sales, Luna forgot the Instagram password.
There is something
fascinating here. AI can analyse extraordinary quantities of information, write
business plans, crunch financial data, generate advertising, communicate with
suppliers and perform tasks in seconds.
But entrepreneurship
isn’t simply a collection of tasks.
It involves
judgement. Instinct. Understanding people. Knowing when the data is wrong, when
an opportunity feels right, when an employee needs support – and when “no
worries” isn’t the appropriate response.
And that instruction
to make a profit?
After four months,
Luna had lost US$62,000.
Perhaps entrepreneurs
can breathe easily for a little longer.
Although, given how
quickly AI is improving, I wouldn’t become too comfortable.
Mathew
Dickerson