After the Component Boom: Where a Precision Machine Shop Fits

1. We are standing at a strange moment

For the past two years, anyone in precision machining has felt the same thing: the work suddenly increased, and it suddenly got harder.

More work, because the vehicles are multiplying. Humanoid robots, quadruped robots, drones, low-altitude aircraft, commercial space, satellite internet terminals, embodied AI devices of every kind — each new vehicle is a whole new set of precision part requirements. A humanoid robot has thirty-odd joints, and every joint contains a housing, a shaft assembly, gears, brackets. A single satellite carries hundreds of precision structural parts.

This is not common in manufacturing history. For twenty years, precision machining demand came from consumer electronics and automotive — a round of scale within a single category. This round is different: many categories exploding at once, and none of them has settled on a technical route yet. Today’s design may be thrown out next month.

The difficulty lives in the same place. Many categories, small batches, fast revision, and tolerances still tightening — four things happening simultaneously, which for a traditional machine shop means running against the grain on all four. What a traditional shop is good at is precisely high volume, stable, repetitive work.

So the question becomes: in this era, what is a precision parts shop actually supposed to do?

This is not a product pitch. It’s a serious attempt to think that question through.

2. First, an uncomfortable prediction

A lot of people want to know whether AI agents will replace human work.

My view: they will replace some of it — but what gets replaced is not “people”, it’s a particular kind of human capability.

Here I want to borrow a Taoist frame. In Taoist internal alchemy there is a pair of concepts: yuanshen (元神, the primordial spirit) and shishen (识神, the acquired consciousness).

As The Secret of the Golden Flower describes them: yuanshen is the innate, original self, and it favours stillness. Shishen is the knowledge and consciousness accumulated after birth, and it favours activity. What’s interesting is how the old texts appraise shishen — it isn’t treated as a bad thing. The commentary says:

“Shishen can enable a person to accomplish great things. But when shishen becomes over-excited, it is the source of the seven emotions (joy, anger, sorrow, thought, grief, fear, shock). It disturbs yuanshen, and this leads to a loss of self-regulation.”

That sentence deserves some thought. It grants that shishen can accomplish things, and in the same breath points out that an over-active shishen dissipates yuanshen.

(What follows is my own extension, not the original Taoist meaning — but I think the extension holds.)

I have come to think of yuanshen as the consciousness that can link to everything.

You might ask: doesn’t the text say yuanshen favours stillness and shishen favours activity? So how can yuanshen be the active one?

My reading is this: yuanshen does not live by moving. It lives by connecting.

The I Ching has a line: “Silent and unmoving, yet it responds and penetrates all things.” The stillness exists precisely so that it can connect with everything. The stillness of yuanshen is not deadness — it is not being filled up by your own thoughts. It stays empty, and so it can hold the whole world.

That is where the two diverge:

• Shishen is consciousness that turns inside the existing individual — comparing, calculating, defending, chasing. All of its material is “the experience I already have”.
• Yuanshen is consciousness that can cross the boundary of the individual — empathy, intuition, resonance. Its material is “the link between me and everything else”.

So yuanshen looks still, but it is the most active connection point there is. It doesn’t produce thoughts. It is the field in which thoughts can occur.

And what is creation, at bottom? It is connecting two things that weren’t previously connected. The steam engine joined to the textile loom produced the Industrial Revolution. A battery joined to a wheel produced the electric car.

Connecting is exactly what yuanshen is for. So saying creativity is rooted in yuanshen isn’t a stretch — yuanshen supplies the capacity to connect; shishen supplies the technique to make the connection real.

Apply that frame to AI agents and it fits neatly: agents excel at classic “shishen-type work” — calculating, optimizing, brute-forcing, repeating, chasing efficiency. They don’t tire. They have no moods. They can “accomplish things” to an extraordinary degree.

What they cannot do is the other category: judging by instinct when information is incomplete, weighing trade-offs by experience when constraints conflict, choosing by taste when there is no right answer on the page.

That is a “yuanshen-type” capability. It isn’t knowledge — it’s judgment, and judgment is built on connection.

My prediction: once shishen-type work is replaced at scale, people will be pushed back toward the yuanshen position. Not because we choose to go back, but because staying where we are will no longer be worth anything.

3. The release of creativity is the real upside

I don’t accept the claim that AI agents will stall society. But I don’t think they automatically produce anything good either.

The likelier outcome is a world where everyone creates.

The logic runs like this. In the past, if you had an idea, a massive wall stood between the idea and the product: tooling, prototyping, pilot runs, supply chain, capital. That wall kept the great majority of human creativity outside. Many people are not short of insight — they are short of the means to turn insight into a physical object. They’re busy earning a living, busy running. Things that should have grown got pressed down.

The wall is getting shorter. Design tools have AI assistance, engineering validation costs are falling, the threshold for small-batch production is dropping. Someone who understands a use case but not manufacturing now, for the first time, has a route to making the thing.

What does that mean for a machine shop?

It means our customer base changes. We used to serve large customers with a complete engineering team. Increasingly we will be approached by small teams and individual creators with an idea and no manufacturing capability at all.

That is a genuinely new proposition for us: can we catch these people?

4. Two things we need to do

One: build links with everyday creators

The traditional machine shop order flow is: customer has a mature drawing → we quote → we produce. That flow assumes the customer already understands manufacturing.

Creators don’t. They may understand their use case beautifully, but the drawing they produce cannot be machined — walls too thin, tolerances specified unreasonably, wrong material, no consideration of fixturing datums.

If all we say back is “this can’t be made”, we lose this customer group permanently — and we lose the era with it.

So the capability has to move upstream:

• Review the drawing early. Not finding fault after they’ve finished drawing, but telling them at the start that “designed this way it costs three times as much; change this and it halves.”
• Lower the psychological barrier to a first order. No minimum order quantity. One piece is fine. A creator’s first piece is usually a single piece.
• Explain the process knowledge. What a machine shop is actually worth is not its equipment list. It’s the mistakes it has already made — which material distorts, which geometry can’t be held in a fixture, which tolerance is money burned for nothing. Put that knowledge out, and the right customers find you.

The writing we do in our knowledge base is exactly this work: letting people who don’t know manufacturing get into a conversation with us.

Two: build internal capability — specifically the part that can’t be described by an algorithm

This is the part I want to stress harder.

When most people talk about building capability, they mean buying more expensive equipment and expanding the floor. I don’t think that’s the key. Equipment can be bought with money. So can anyone else’s.

The real internal capability is the stuff that cannot be written into a work instruction and cannot be brute-forced by an algorithm.

An example. A thin-walled joint housing, 1.5 mm wall. A three-jaw chuck distorts it on contact; release it and the part springs back; the measurement passes on the machine and it’s out of tolerance the moment it comes off.

How do you solve it? Swap to a vacuum chuck, change to axial clamping, split roughing from finishing, insert a stress-relief anneal in between — is there a standard answer to that? No. It depends on what the part looks like, what state the material is in, the temperature in the shop that day, even the mood of that particular machine.

This kind of judgment is beyond AI today, and I expect it will remain so for a long time. It isn’t knowledge retrieval. It’s the application of experience.

This is the real moat a traditional machine shop has in the age of AI agents.

So my definition of “building internal capability” looks like this:

That last row matters most. Holding ±0.005 mm on one part — plenty of shops can do that. Holding it across five thousand parts in a row is a different trade. The first comes from skill. The second comes from a system: process data, SPC, and material that can be traced back to the heat number on every batch.

An agent can optimize a process. It cannot take responsibility for a batch’s stability on your behalf.

5. Why “fix yourself first” is the only road

There’s a line from The Great Learning: “Cultivate yourself, regulate your family, govern your state, bring peace to the world.”

It usually gets read as moral instruction. I think it’s an extremely practical order of operations.

Read it for a manufacturing business:

• Cultivate yourself — build process capability to the point where nobody can substitute for it. This is the foundation.
• Regulate your family — lead your team and your supply chain well. However strong one person is, one person cannot produce a stable batch.
• Govern your state — take your position in the industry ecosystem. How many creators you can catch is the measure of your value.
• Bring peace to the world — only then are you in a position to take part in the era.

The order cannot be reversed.

And this is my final answer to “what should we do in the age of AI agents”:

Don’t panic. Don’t chase the trend. Do the thing in front of you to a standard others can’t reach.

The wave of creators is coming, and it will be large. But a wave only lifts people who are ready. It doesn’t hold up anyone who can’t swim.

6. In closing

We are a precision parts shop. 23 years in manufacturing, 15 years in CNC precision machining, 30 machines. We make joint module housings, reducer planet carriers, precision structural parts — the kind that look unremarkable but fail if they’re slightly off.

The question this era is asking us:

As vehicles multiply, as creators pour in, as agents take over calculation and optimization — what value does a machine shop have left?

My answer is plain:

What’s left is judgment.

What’s left is the judgment that can’t be described by an algorithm and can only be ground out by hand over years. The judgment to take on a job when the drawing is incomplete; to know which tolerance to protect when two of them conflict; to know how to trade cost against performance.

That judgment is what we stand on in this era.

And turning that judgment into capability that helps more creators — that is the whole of what we intend to do next.

The views here are the author’s own. The reading of yuanshen and shishen is the author’s extension, not the classical Taoist definition. Fellow manufacturers are welcome to push back.