AI, Boredom and Business Efficiency

Excuse Me

We’re in a bubble of comment about AI bubbles. In fact, the bubble may have burst, as online searches for “AI bubble” are in retreat. This did not stop Jeff Bezos being the latest big name to use the b-word.

Derek Thompson summed it up in a soundbite. Annual expenditure on AI is the size of Singapore’s economy. Annual revenue from AI is the size of Somalia’s. That’s a 42x difference.

Bezos made the point that overinvestment is worth it in time. Society as a whole will be the winner. Last I checked that’s not why companies are in business.

Last week I discussed why we expect too much from technology in the short term. Our stimulus-driven, reptilian brains want it all now. We’re just not good at waiting. By the time technology transforms society we’ve gotten bored.

Boredom is a problem. It dulls our sensors and we make mistakes. Alleviating boredom is what a lot of technology is about. Even if all it does is occupy hours with mindless streaming slop.

The big concern is that AI replaces us. While the internet turned our minds to mush, AI is far worse. It will leave us with nothing. Just a vast emptiness where our boring jobs used to be.

Excuse me if I struggle to see free time as a problem.

Rise and Fall

We’ve been busy at work of late hiring for new roles. While we automate business processes for clients, we also provide people. Big companies want to do their own automations. They need specialist skills to help out for a while.

Hiring is a numbers game. The more candidates you source, the better and faster you can fill a role. This matters when clients come to you only after they’ve exhausted their internal resources. Every request is urgent.

Urgency is not an excuse to skimp on customer service. Presentation becomes more important when you act at speed, because it is a strong signal that you remain in control. Typos and poor formatting indicate you’re wobbling.

Processing CVs is time-consuming. Resumes need appraising and repackaging in a standard format. The whole point of filtering candidates for the client is to take the effort out of reviewing. The handpicked CVs must be easy to compare. They need consistency.

This is a challenge when every candidate wants their CV to standout. Formatting lots of CVs takes days. With an increased workload we need more people. Yet we don’t know how long the current boom will last.

You can run CVs through ChatGPT but details get lost. The rules are applied inconsistently. Formatting such as bold, italics, bullet points and branding still needs the human touch.

This is not what big tech wants. The idea behind that Singapore-sized spending is that LLMs will do everything for us. If too much prompting is required for results, this limits the potential market. The model-makers are determined that GPTs will be powerful enough that anyone can use them.

Yet you don’t need a huge brain to reformat CVs. You require an eye for salient details and a consistent approach. AI can do this at scale, 100 CVs at a time. But not out-of-the-box.

Our team has built ReadyResume. Drag and drop files, input role and seniority, and the system converts the CV to a branded template in Word or PDF.

We used a tool we’re trying called AugmentCode. It understands our codebase and the context of the project. Copilot is fine for everyday speed in single files. System structure requires something else. We built, tested and launched in under a week.

ReadyResume is tailored to our needs. It can be used with a few tweaks by recruitment consultants and companies making many hires, and we already have enquiries. These businesses understand there is an extra step involved in getting exactly what you want.

The AI assistant helps us keep track of the codebase. That way if people leave, the knowledge stays with us. We’re avoiding tech debt by not creating it in the first place.

The biggest benefit however, is that we can scale up our business. We do with this without hiring people who we may not need for more than a few months. Our AI spending will rise and fall instead.

A Bankable Benefit

As AI is new, no one knows how much they should be spending on it. Over time the biggest cost will be running models. Building an efficient system on a range of models is the best way to limit your liabilities.

It’s worth spending a little more today to design a system that scales. Vibe coding your DIY tools is all very well, but that’s not how to build a business.

For example, solopreneurs might cook up a process that they giveaway, but the business model remains their paid services. Bigger businesses need more than free tools. They require results.

Right now that means designing systems to remove pinch points in your business. The best way to identify these is where there is demand for more hires.

The resulting spending won’t bridge the gap between Singapore and Somalia for the AI industry. But it will be a bankable efficiency for your business.

Questions to Ask and Answer

1. Where in your business are you always short of time or resource?

2. What is the blockage in the process?

3. What repetitive task can you automate to remove this blockage?