The Value Creation Protocol

Part III — The Core Framework Traps

Chapter 7

The AI Replacement Trap → Value-First AI


When automation fights augmentation, both humans and organizations lose

In June 2025 I hosted ten live conversations, one for each of the ten Core Framework Traps. The first was this one, with Nico Lafakis. Nico and I do a daily show together, Value-First AI Daily. Here is what he said people were hearing:

"The replacement trap is, I think, what everybody is probably been hearing in the last two weeks. It's a drum that wasn't beat very loudly last year. It was tippity-tapping around Christmas … and that is your job is going to get replaced by AI."

If your people can hear that drum, it doesn't matter much what your AI plan says on paper. They'll work with the machine the way anyone works with the thing that's coming for their job. Carefully, and with the best of what they know kept to themselves.

What this is

The AI Replacement Trap is one of those ten. Its definition:

Believing AI's primary purpose is to replace human workers rather than enhance their capabilities.

Nobody walks into this on purpose. I think the habit comes from the industrial age, when a machine that could do a job usually meant fewer people doing it. The trap is carrying that picture over to thinking work, where what your people know about your customers and your business is exactly what the machine doesn't have.

It runs against one of the Five Core Beliefs, AI-Human Partnership over Replacement: "AI should multiply human capability, not replace it." Chapter 2 gives that belief in full.

The way out has a name too: Value-First AI. Its shift is from one question to another. From "How many people can we eliminate?" to "How much value can each person create?"

That second question needs the word value to mean something exact, so here is the definition these books use:

What a person or organization can now do that they could not before. Not what they were sold, not what was produced for them, not a promise about the future — value is the real change in what someone can do, and only the person who received it can say it happened.

The Value Era makes the case for that definition in its Chapter 14, "Only the person who received it". For this chapter you only need to set it beside the phrase "time saved" and notice the two aren't the same thing. Time saved is what got removed. Value is what someone can now do.

How to tell where you are

Start with your ears. In the article I published in September 2025, I wrote down three sentences I was hearing constantly:

"AI saves me 30 minutes a day."

"This tool cut my writing time in half."

"Now I can process twice as many leads."

Every one of them measures the tool by what it took away. Not one says what anybody can do now that they couldn't do before. If those are the sentences in your status updates, you're standing in the trap, even if nobody has said the word "headcount."

Then look for these, this week:

  • The business case for an AI project is written in positions or hours removed.
  • The people who do the work heard about the project after it was chosen. They were asked to test it. They weren't asked to shape it.
  • The first thing you automated was a process everyone already complained about. Now it's a faster process everyone complains about.
  • When AI comes up in a team meeting, the questions are about job security, or there are no questions at all.
  • Your most experienced people have gone quiet about how they actually do what they do.

There's a quieter version where nobody is afraid and the trap is still running. Nico again:

"It's this trap where you think that it will be creative for you and it will figure out how to apply the creativity for you. And that's where it starts creating this huge gap and this huge mistake."

That's replacement pointed at yourself. You hand the machine the thinking and wait for it to come back finished.

One question sorts all of it. It's from the same 2025 article:

"Does this AI implementation enable me to contribute something that wouldn't exist without it, or does it just help me do existing work faster?"

Ask it about the last AI project your team shipped.

What to do

  1. Change the question before the next AI decision. Take "what can it do in place of our people" off the table. Put these three up, in this order:
    • "What value am I uniquely positioned to create as a human being? As a team? As a leader?"
    • "How could AI partnership enable me to create more of that value?"
    • "What entirely new forms of value could AI partnership make possible that don't exist today?"
  2. Say out loud what the project is not for. If the project isn't aimed at anyone's job, say that in plain words to the people it touches. If you can't say it truthfully, stop here. That's your finding, and the rest of this list won't work until it changes.
  3. Find where coordination is eating someone's judgment. Sit with one person and go through their week. Mark the chasing, the copying between systems, the assembling of updates, the reminding. That is the machine's share. What's left is creativity, judgment and connection, and that's the share you're trying to give them more room for.
  4. Put the person who knows the work in charge of the machine's work. They bring the idea. They hold the standard. They're the one who says "no, that's wrong." Nico's warning fits here: "You can't just pick up these tools and throw them at the wall with an idea and expect it to come back with the most amazing thing ever."
  5. Start with one piece of work where something new becomes possible. Look for work that couldn't happen before. The same work done faster doesn't count. My own first proof was small. Earlier in 2025 I was live streaming and could never take the next step to repurposing any of it. With an AI toolset handling the transcription, the editing coordination and the distribution, I was producing two daily podcasts that wouldn't have existed without it. I wrote later that year: "This isn't time-saving. This is value expansion."
  6. Decide how you'll know before you begin. Write down who is supposed to be able to do something new, and what it is. The next section is how you check.

How you know it worked

You ask the person whose work changed, and you ask it plainly: what can you do now that you couldn't before?

You're listening for an answer in their own words that names something specific. Something they make now. A conversation with a customer they can have now. A problem they used to pass along and now finish.

Two cautions about what counts.

Who is speaking matters. When the vendor, the project lead or you report that the team saved hours, that's the provider taking credit for what was removed. When the person says "I got time back, and here's what I did with it," that's the one who received it, naming it. Same words, different speaker. Only the second is evidence.

A feeling is worth writing down, as a feeling. If someone says "that felt faster," keep it. It's what they experienced. It is not a measurement of time saved, and it shouldn't be reported as one.

What people do counts as much as what they say:

  • They bring you the next idea without being asked.
  • Their questions change from "do I have to use this?" to "can it also do this?"
  • The experienced people start explaining how they decide things, because now there's somewhere for it to go.

A count of logins, prompts or automated tasks tells you the tool is being touched. It can't tell you any of this.

Where it goes wrong

The Measurement Trap is the one beside this (Chapter 14). It usually starts the day the project needs a number for a board slide. "Hours saved" is the easiest number to get. Once it's the headline, somebody does the next piece of math: whose hours, and how many of those people do we still need? You're back at the first question with a nicer chart.

Other places it turns:

  • The vendor's pitch arrives in replacement math. Your business case inherits it line for line.
  • You automate a broken process. The dysfunction now runs at machine speed. Of the twelve traps, this is the one that threatens to accelerate all the others.
  • The tools get chosen and built apart from the people who do the work. You end up with two groups: the ones who run the AI and the ones it's run on.
  • Partnership is the slogan and next year's budget still assumes fewer people. People read budgets.

Read further

Stop 1 of 6: What this is

Back to contents