Do not train your AI replacement for free

As a senior software engineer, my brother-in-law knows more about the AI economy than most.

He learned the tools, helped build and train his company’s AI agents, and became one of the core people leading the transition.

Then, the healthcare software company where he worked reduced its full-stack engineering team from 60 people to eight.

He was one of the 52 who got laid off.

Only the company knows the exact reason for the mass dismissal, but I bet it has something to do with this new reality -- AI can help a smaller team do the work of a much larger one.

That leaves employees with an uncomfortable choice. Help improve the system and risk making their own roles less necessary, or refuse and risk being pushed out even sooner.

My answer is not to reject AI.

It is to refuse to train your digital replacement for free.

A software factory takes a product request and moves it through a series of AI agents. One breaks the request into tasks. Another studies the codebase. Others write the code, run tests, correct errors, and prepare the work for review.

The humans, meanwhile, set priorities, handle the difficult exceptions, and decide what ships.

Sure, this isn't exactly a brand new concept.

GitHub already lets companies assign an issue to a coding agent that works asynchronously. Later on, the project supervisor can pull and review the output. Boom. It's that simple.

And it reminds me of one of my favorite movies: Office Space.

Remember when the two Bobs took days to decide who was gonna get shit-canned?

Well, it's not 1999 anymore.

Today, the Bobs are redlining org charts based on what the tech stack can now do post AI implementation.

The people running major companies have become remarkably direct about where this leads.

Want me to keep going? I have like, 72 more of these. ^^

Basically, companies once had to justify automation. But now, employees are being asked to continually justify their existence.

American workers have seen a version of this bargain before.

When Fruit of the Loom moved production from Texas to Honduras, the company sent veteran employees overseas to train the lower-paid workers who would replace them.

The software engineer faces the same problem, but under a different mechanism. The work stays within the company, while automation reduces the amount of human labor required to complete it.

The common thread here is the knowledge transfer.

That employee has the most leverage before the transfer is complete.

Don't get me wrong -- refusal without leverage is just a resignation letter with extra steps.

That's why, in May, 2,100 University of California IT workers voted to unionize, citing AI governance. They joined the CWA, the current largest tech union in the country.

Their demands? Layoff protection, transition plans, the right to reject unethical AI work, and a share of the productivity gains.

Collective bargaining cannot guarantee that every job survives. A company can outsource the project, accelerate the rollout, or walk away from negotiations.

But in my opinion, that would be dumb.

Everyone else is already negotiating over "AI savings."

A recent contract presented to my company would require us to disclose where AI appeared in the work and give the client the right to reduce the price. The potential client believes that a cheaper production process should change the economics of the deal.

Employees have grounds to make the same argument.

When AI creates enough value to reopen the price with a partner, it creates enough value to reopen the terms with the people powering those engines.

No, I am not suggesting the Sarah Connor plan for a Skynet takedown.

We can negotiate at least one round before blowing up the data centers.

(Please don't blow up any data centers...)

Our Chief Strategy Officer and former Ourisman COO, Jack Ballinghoff, told me a story about Sonic Automotive CEO Bruton Smith today.

He said he once watched Bruton walk into a room full of dealers and say, "The auto industry has created more millionaires than any other industry in the history of the United States. And it will continue to do so."

My point here is that a dealership has historically offered a rare path to serious economic mobility.

Someone can enter the biz without a four-year degree, learn from the ground up, and eventually run a department or an entire store. Hell, maybe they'll even end up owning one.

But AI is putting the current headcount at stake. A lot of times, it is removing the early assignments through which people learn the business, earn trust, and become valuable enough to move up.

A customer may happily let software schedule an appointment, generate a trade estimate, or send a repair order. But the moment the estimate includes a $6,000 repair, the customer wants someone who can explain it, defend it, and put a name behind it.

CDK Global recently found that 80% of dealership service customers said their advisor had earned their trust by the end of the visit. And the crazy thing is, CDK also found that changes in advisor communication and follow-through could swing Net Promoter Scores by 40 to 60 points.

That relationship carries the sale, the service visit, the recovery from a mistake, and the customer’s decision to return three years later.

AI can handle the administrative tasks in that relationship, but it can also remove the relationship layer itself.

That is where the dealership employee ends up in the same bind as my brother-in-law.

Companies will keep asking employees to document their inputs, correct the outputs, and turn years of judgment and relationships into software. Workers cannot stop every rollout, but they can stop treating the transfer of knowledge like ordinary housekeeping. They can negotiate for a role in what comes next, protection if the team shrinks, and a share of the value they helped create.

New at Freddy Media

We've just signed Amalfi Jets as a media client, the experts in private, luxury travel.

And if you've spent any time on TikTok, you've likely seen their deals get made in real time. That content engine helped build a $100 million-plus private aviation brand and a digital magazine covering the entire luxury travel world.

Want your brand in front of this audience? Hit reply to this email.

Freddy Media would like to share our best practices with you

Here are some products we find useful, and would encourage you implement in your company.