When you think of an "AI factory,"
the first image that might come to mind is:
humanoid robots walking along production lines.
robotic arms working at high speed.
almost no humans in sight.
The whole factory looking like something out of a sci-fi movie.
But the real question is:
What about the existing old factories worldwide?
The machines inside may have been running for 10, 20 years,
or even longer.
Factory owners can’t simply tear down and replace multi-million dollar equipment overnight just because AI has arrived.
This is the challenge that US startup Harmoni aims to solve.
They didn’t start by building a humanoid robot.
Instead, they did something that looks surprisingly ordinary:
They attach a tablet right next to an old machine.
But the truly important part
isn’t just the tablet itself.
It’s that this tablet is adding a new AI layer to the “old factory.”
Harmoni Secures $10 Million in Funding
On September 9, 2026,
Harmoni announced the completion of a $10 million Series A round,
led by Bessemer Venture Partners.
The company also unveiled their new manufacturing AI:
HAL.
Harmoni positions itself as a Factory Orchestration platform.
In simpler terms,
it's not just adding another AI chatbot to the factory.
Instead, it aims to:
Reunite factory information scattered across various places;
For example:
Which work order is currently in production?
Which part is being made?
Which machine is running it?
Where are the engineering drawings?
What’s the correct processing procedure?
How far has quality inspection progressed?
Where are problems occurring?
Previously, such data was often dispersed across:
ERP systems.
Machine control systems.
Paper documents.
USB drives.
Engineering department computers.
On-site workers’ experience.
Even residing solely in the mind of a senior technician.
Harmoni aims to connect all these dots first.
Only then can HAL understand what’s happening.
The Tablet Attached Magnetically Garners More Attention than the AI Itself
Business Insider reported on a very specific design choice.
Harmoni installs customized tablets beside factory machinery,
which can be magnetically attached and fixed in place.
Their solution doesn’t only serve the newest smart machines.
The report notes that Harmoni’s equipment can connect to machines ranging from cutting-edge devices
all the way back to even WWII-era equipment.
This image is crucial.
Because it shows that:
AI factories don’t have to start from scratch.
You can have a very old processing machine,
add a modern tablet next to it,
then connect work records, machine data, ERP, documents, and on-site workflows together.
The machine itself doesn’t suddenly become AI-enabled.
But through this new system, workers can more easily understand:
What the machine is doing now.
What needs to be done next.
Where issues might be arising.
Workers Previously Had to Repeat Tasks in Multiple Places
Business Insider mentioned one of Harmoni’s clients:
WessDel,
a company that manufactures precision parts for aerospace and defense industries.
Before using Harmoni,
workers might have to queue up at different kiosks to clock in,
then manually input machine information.
A WessDel manager said
this kind of administrative work could consume almost an hour each day.
Now, workers can simply swipe their ID cards on Harmoni tablets next to machines,
and the system automatically logs machine activities.
This may not sound like an “AI revolution.”
It might even seem mundane.
But this is precisely the point.
The real drain on factory efficiency is often not:
“machines can’t think.”
Rather, it’s:
humans spending time searching for data, re-entering info, verifying versions, queuing, making calls, and asking around.
If AI can first eliminate these time sinks,
it might prove more valuable than simply having a robot walking around.
Why Is HAL Different from Ordinary Chatbots?
Imagine placing a generic conversational AI directly into a factory.
A worker might ask:
“Why is this machine running slower today?”
The AI’s first questions would likely be:
Which machine?
Which part is it producing now?
What is the normal speed?
How was the last batch?
What are the engineering requirements?
Where are the maintenance records?
Because it has no understanding of the on-site context.
Harmoni operates differently.
Their platform already knows:
What job is currently running.
Which part is being processed.
Which machine is in use.
Which process is active.
Which engineering requirements must be met.
All this context is provided directly to HAL.
So workers can simply ask:
“Why is this machine slower today?”
Without having to type out extensive prompts.
This is what makes HAL truly interesting.
AI Must First Know “Which Machine You Are Standing In Front Of”
When we normally use ChatGPT, Gemini, or Claude,
we often have to provide background information, such as:
“This is a client contract.”
“These are our company policies.”
“These are yesterday’s data.”
Because the AI doesn’t know our current context.
But manufacturing work is highly location-specific.
A person standing in front of Machine A
and a person standing in front of Machine B
will have entirely different questions.
Producing part X today,
and part Y tomorrow,
may have different standards.
So what manufacturing AI really needs is not just:
“a smart model.”
But:
It must understand who the person is, which machine they’re by, and which work order is underway.
This is why Harmoni emphasizes context-aware AI.
Workers Can Even “Talk to Machines” Directly
Business Insider describes how HAL allows on-site workers to ask questions directly,
request maintenance tasks,
or leave issues for the next operator.
Harmoni co-founder David Caputo calls this:
enabling workers to truly “talk to the machines.”
Don’t misunderstand.
It’s not that the old machines suddenly grow mouths.
Rather, HAL integrates:
machines,
work orders,
engineering data,
past production records,
and on-site workflows into one unified context.
So a worker’s natural language can now be
converted into actual actionable factory information.
Why Not Replace Everything with Humanoid Robots?
Because the real world isn’t that simple.
Harmoni itself points out that
traditional automation excels at handling highly repetitive tasks.
But many manufacturing environments still face:
frequent line changes,
small batch varieties,
on-the-fly anomalies,
human-machine collaboration,
complex engineering requirements,
and coordination between different systems.
These challenges won’t disappear just by installing a robot.
More realistically,
many small and medium-sized factories haven’t fully digitized their basic data.
Shokoufeh Mirzaei, a manufacturing engineering professor at Cal Poly Pomona, told Business Insider,
that many small and medium US factories still operate on very old legacy systems.
Businesses aren’t always ready to replace expensive equipment entirely.
But they might accept
a retrofit approach —
modifying existing equipment instead of rebuilding.
It’s Like Upgrading an Old House to a Smart Home
Imagine living in a 30-year-old house.
You want to turn it into a smart home.
The most extreme approach is to:
tear down the house,
rebuild,
rewire everything, and
replace all equipment with new devices.
An alternative is to:
install smart locks first,
add sensors,
replace a few controllers, and
slowly connect the existing equipment into one system.
What Harmoni is currently doing for factories
is much closer to the latter.
They aren’t saying:
“Your old factory is useless.”
Instead,
“Let’s see if it can live through one more AI generation.”
This Is More Likely to Scale Before Humanoid Robots
Humanoid robots are, of course, very appealing.
They make for impressive visuals.
A robot walking into a factory,
carrying boxes,
picking up tools,
operating machines —
anyone who sees this knows:
the future is here.
But the countless factories worldwide that are truly producing parts, food, equipment, and materials
will not suddenly vanish just because of new technology.
They already have machines,
existing workflows,
workers,
ERP systems,
customer specifications,
certifications,
and decades of established practices.
So when AI truly enters manufacturing,
it’s more likely not to be:
“a new world replacing the old one entirely.”
But rather:
adding a new layer of intelligence over the old world.
This Changes How We Imagine AI
Many people, when they think of AI,
ask first:
Can it replace humans?
But Harmoni’s case shows a different path.
Their product logic focuses more on:
reducing the time workers spend searching for information,
minimizing repetitive data entry,
reducing queuing,
finding the correct documents faster,
quickly diagnosing machine issues,
and helping the next operator understand what happened in the previous shift.
Harmoni even claims that after automation,
each operator’s productive time can increase by over 200 hours per year.
This figure is based on their client data,
not an independent industry average,
so it doesn’t mean every factory will see the same results.
But it points to one direction:
AI’s value isn’t always in performing flashy new tasks.
Sometimes, it’s simply about:
taking back those wasted 10, 20, or 30 minutes every day.
AI Competition in Manufacturing May Split into Two Paths
One path is:
building new AI factories.
Highly automated.
Many robots.
AI-ready from the design phase.
The other path is:
retrofitting the hundreds of thousands of existing factories.
Connecting old machines to new systems,
linking people, equipment, and data anew.
Harmoni chooses the second.
Business Insider cites estimates of about 280,000 existing factories in the US.
If a large number of machines won’t be replaced anytime soon,
then “how to bring old machines into the AI era”
could become a massive market on its own.
So the Real Story Isn’t the Tablet
From a distance,
Harmoni’s story might seem unremarkable.
No humanoid robots running around.
No fully automated dark factories.
No sleek, futuristic robotic arms neatly lined up.
Just:
an old machine,
a worker,
and a tablet by its side.
But perhaps this approach is closer to how AI will truly enter the real world.
Technological revolutions rarely wait for all the old things to vanish before starting.
More often,
new technology becomes widespread because it finally learns:
how to work alongside the old world.
So the question worth asking today might not be:
“When will humanoid robots take over factories?”
But rather:
Before that day arrives, can AI help us better understand, manage, and reduce wasted time on a machine that’s been running for 30 years?
This could be the first step toward faster AI adoption in manufacturing.
Today, make a little progress with AI.
Learn an AI trick each day.
Save some time every day.
Improve a little bit every day.
SasaDaily, growing with you.
Recommended Reading
Why Companies Struggle with Legacy Systems Despite Having the Best AI
