Hey there,
Earlier this week, I presented our Q4 priorities to the Lupa team and focused a lot on quality.
It’s not only about how much we get done or how fast we work. What really matters is the quality of the work.
One point I kept coming back to was AI. We make continuous use of it. It can accelerate research, help analyze information, automate everyday tasks, and move us from an idea to a useful outcome much faster.
But I don’t want it to replace our thinking.
The standard calls for careful work; you have to understand the customer, look at the evidence, and decide whether what you’re producing actually makes sense.
The next day, Anthropic released a study with numbers that helped explain why I see AI this way.
Researchers estimate that robots can already handle 74% of physical work tasks in the U.S., at least in some situations. Add in tasks managed by large language models, and about 80% of working time in the U.S. now involves some form of technology.
But here’s the key number: right now, robots are only as cost-effective as human workers for 0.3 percent of jobs.
We need to get better at telling the difference.
Just because something is possible doesn’t mean it makes financial sense. A job ad doesn’t always mean someone plans to hire. And a country’s unemployment rate doesn’t show what kind of talent is actually available.
This week’s three stories all highlight that gap.
Let’s get into it.
🌐 News Shortlist
1. Brazil banned online betting. The skills built there aren’t disappearing.
Recap: On September 25, President Luiz Inácio Lula da Silva approved a provisional measure that prohibited fixed-odds betting throughout Brazil, including sports and online casino games. The rule immediately halted platforms from taking in new bets, and customers have until October 5 to withdraw their balances; authorized betting websites and apps are set to go offline starting on October 6. Although the measure currently has the force of law, it needs Congress’s approval within 120 days to remain in effect.
Timing played a role here. Lula signed the bill just nine days before the first round of the Brazilian presidential election, and both his supporters and opponents have talked about what it means politically.
But the issue also became politically important for a personal reason: Brazil has a serious gambling problem.
The federal government says that over 1.3 million Brazilians have asked to be excluded from betting websites, with about one-third of those requests made during this year’s World Cup.
These are people who have chosen to ask the government to stop them from gambling.
I understand why there’s a need to protect families from addiction and financial trouble. But we should also pay attention to another group: the people who built this industry.
Brazil had established one of the world’s largest online betting markets, and before the ban, the regulated sector included 85 licensed operators across 185 platforms, according to reports based on government data.
These companies required engineers who could build high-volume platforms. They needed payments teams that could handle huge sums of money. They needed fraud and risk specialists to monitor millions of transactions. They also needed compliance staff to deal with the new regulatory system.
They also employed individuals from a range of departments, including product, data, finance, customer operations, and growth marketing. Many of those jobs are now in danger.
CNN Brasil estimated that about 15,500 posts could be affected by the shutdown, with technology expected to be hit hardest. This figure represents an estimated number of losses, not the actual number of layoffs, and the industry’s legal situation is still being contested.
An industry can vanish much faster than the skills it creates. That’s why I keep coming back to AI.
We spend a lot of time debating whether technology will take away certain jobs.
But sometimes the better question is the opposite: when technology, new rules, or economic changes suddenly remove a job, what happens to the skills people had?
Those skills can be used elsewhere. Fintech is an obvious destination. So are e-commerce, marketplaces, SaaS, payments, and other digital businesses.
U.S. companies that hire remotely can tap into this same talent pool.
Advice:
If you’re hiring in product, engineering, payments, fraud, compliance, data, or growth, add Brazilian betting companies to your sourcing map now.
Don’t treat these professionals as cheap talent. That would be short-sighted and disrespectful.
Reach out to them. Many have spent the last few years solving tough problems at scale.
The industry they worked in is changing quickly, but their experience remains valuable.
2. There are 7.1 million job openings in the United States, but that does not mean companies are actively hiring.
Summary: The most recent JOLTS report was released on Tuesday. By the end of August, job openings in the United States had fallen from 7.335 million in July to 7.079 million.
That’s still a lot of jobs. However, employers made only 5.192 million new hires that month, with quits holding at about 3.1 million and layoffs and discharges at roughly 1.6 million.
At first glance, this seems reassuring. Companies aren’t laying off workers in large numbers, but they’re not hiring aggressively either.
The Associated Press stated that the labour market features a high degree of security for people who have jobs, whereas those who are looking for work are encountering a far more difficult situation.
For years, people have looked at job vacancies to understand labor demand.
A job opening isn’t the same as a hire. One number shows how many jobs are open at a given time, while the other shows how many people actually get hired each month. So, subtracting 5.2 million from 7.1 million doesn’t really make sense.
But when you look at both numbers, you see something important: companies may want talent even if they’re not ready to hire.
Maybe the budget isn’t approved yet. They could be waiting until next quarter. Sometimes the job description asks for a perfect candidate who doesn’t exist. Or they might be interviewing while deciding if AI can handle some of the work.
The role might really be open, but leadership may have decided they can wait to fill it. For candidates, the result is the same: the job exists on paper, but no one gets hired.
This is another example of the Anthropic 74% versus 0.3% issue. A robot might be able to do a task, but that doesn’t mean a company plans to replace the person doing it.
Advice:
Break your hiring dashboard into three numbers—approved roles, active searches, and completed hires. If those numbers are far apart, find out why.
If a job has been ‘open’ for three months with no real progress, stop treating it as an active search.
Either decide to hire someone or close the job. Being clear helps your team, your recruiters, and the candidates who spend hours interviewing.
3. The unemployment rate in Chile is 9.6 percent, but that’s not the figure I’d keep an eye on.
Recap: Chile released its latest employment figures on Wednesday. From June through August, national unemployment was 9.6 percent, up one percentage point from the same period the previous year. Employment fell 0.9% on an annual basis, and the number of unemployed people rose 13%.
That’s the headline. But if you’re hiring in Latin America, there’s another number that matters even more.
Year-on-year employment in Chile’s information and communications sector fell 16.8%.
Now, focus on the Santiago Metropolitan Region. The number of people employed in information and communications work decreased by 25.1%. The number of people employed in formal salaried jobs in the area also decreased by 5.5%.
That tells a very different story than just saying ‘Chile has 9.6 percent unemployment.’
I often see companies make the mistake of treating Latin American countries as if their talent markets never change. Argentina is good for X, Colombia is good for Y, Brazil is expensive, Chile has good engineers.
But labor markets don’t work that way. They’re always changing.
Companies grow and shrink. Currencies go up and down. Entire industries get funding or lose it. Regulations change. Employers in one area might suddenly hire hundreds of people or let them go.
Brazil is a clear example this week. Chile shows a quieter version of the same thing.
A national unemployment rate tells you something about the economy, but it doesn’t show whether it’s getting easier or harder to hire the people your company needs.
We should also be careful about what these numbers don’t show. Just because the number of information and communications employees in Santiago fell by 25.1% doesn’t mean a quarter of the city’s software engineers were let go.
The category includes more than just software, and the INE data shows only that jobs fell, not why. But a drop this big is a good reason to check the market again, since your assumptions from six months ago might no longer hold.
Advice:
At a minimum, check labor conditions by country, city, and occupation quarterly when hiring technical or digital personnel.
If you’re hiring in the region, I’d take another look at Santiago. Look at the response rates of the candidates. Look at what they expect in terms of compensation. Speak to some of those who have most recently resigned from local employers.
You don’t have to predict the Chilean economy. You need better information than the company competing with you for the same candidate.
I keep coming back to that Anthropic study for a reason.
74% versus 0.3%.
The first number shows what robots could do in theory. The second shows what companies are likely to do in practice.
That gap is everywhere.
Brazil might be able to shut down an industry, but it can’t take away the skills thousands of people gained while building it.
The United States could have 7.1 million job openings even if companies don’t make 7.1 million hires.
Chile might have one national unemployment rate, but the labor market in a specific sector in Santiago can change much faster beneath the surface.
That’s also why I’m skeptical whenever someone claims to know exactly what AI will do to hiring.
Yes, AI will automate work. We’re already using it for that. But the real question isn’t just what AI can do.
It’s about the tasks that are now cheaper and faster, but still need human judgment, because people can focus on them now that the repetitive parts take less time.
That’s exactly what I meant when I told my team this week: I want artificial intelligence to speed up our work, not replace human thinking.
The technology is getting very good. Our job is to make sure our thinking improves along with it.
Until next time,
Joseph