Automating yourself irrelevant

Amazing photo by Jeff Eaton

I’ve been selling my brain for money for the longest time and the better I got in doing things, the better I sold, but nothing sold as well as me saying “My job is to make myself irrelevant.” and really standing behind that sentiment.

This is similar to the case of doctors — they make themselves irrelevant by solving your problem and the better ones even try to help you avoid problems with preventive care. You would not want a doctor who would just try to get you back every week and maybe prescribe some more expensive tests by others in order to maximise customer profit?

A company has a few ways of earning money, either having a good product they can sell to many others, or great people that they can place over and over again, but more than often companies expect customer revenue to increase infinitely and even though I’m a greedy capitalist, I don’t agree with the means to getting to that end most of the time.

During my agency years, the goal was to get in the customer, do a decent enough job and then hold the fort. Sell up, sideways, North and South and anytime you can, very rarely you went up there to do a single thing and leave, instead “losing a customer” was considered the largest failure — and for most cases, it is true. It’s much easier selling to an existing customer instead of finding a new one, but any company that wants to maintain their position in their customers permanently cannot really commit to automation as their main goal, because the point of automation, the dream, is to make everyone irrelevant and unemployed, even us.

Naturally, I know that perfection does not exist, especially now, especially with automation of corporate systems and data. But it is doable at least in theory and the more we teach deep learning computers, the less human intervention we need.

I spoke with one of my accountants today and he had an issue with calculating the salaries of 800 people. Incomplete forms, late fillings and special tax cases rendered it “impossible” to handle automatically which caused me to stop him immediately and open up the situation. Each element that was logical enough to divulge was also a clear case of automation — again, not perfect 100% automation, but by just handling every certainly correct one automatically and sending every incomplete item to a human, whose usage would in turn teach the system to recognise the action from that error, allowing the machine to handle itself next time — repeat ad nauseam until you have 50–60–70 percent automation which allows one person to do the work of five people.

Realising this point of automation, meaning that everyone at Razormaker has committed to this cause of making ourselves irrelevant within and without, inside and outside in everything we do has put us on a track that’s impossible to veer from. It will take time, but it’s a path that’s easy to follow but hard to reach the goal.

Source: Deep Learning on Medium