Kinetica
Menu

AI that moves the real world.

Kinetica follows artificial intelligence that’s already on the job — on factory floors, in warehouses, in hospitals and on the grid. We care about what runs every shift, not what looks good on stage. Implementation over demos.

See AI in the field Read the thesis
Welding cell
Pick and pack
Quality inspection
Route planning
Patient intake
Grid balancing
Crop scouting
Shelf scanning
Site survey
Call answering
Welding cell
Pick and pack
Quality inspection
Route planning
Patient intake
Grid balancing
Crop scouting
Shelf scanning
Site survey
Call answering

Most AI lives in slide decks. Some of it already runs every shift.

The loudest AI stories are about what might happen someday. The more interesting ones are quieter: a robot that hasn’t missed a shift in months, a vision system that catches defects before a human could, an agent that answers the phone at three in the morning. Kinetica exists to put those stories in front of people.

Now: AI at work

Robots, vision systems and agents deployed inside real companies and industries today. Measured by uptime and output, not by how good the demo looked.

Next: AI in abundance

The chips, models and infrastructure that keep pushing the cost of intelligence down — until AI is cheap enough to run everywhere, for everyone.

Demo versus deployment

The same technology can look identical in a video and behave completely differently in a factory. This is the difference Kinetica pays attention to.

Runs for

A demo A few minutes, on camera
A deployment Every shift, every day, for months

Measured by

A demo Views, applause and likes
A deployment Uptime, output and cost per task

Environment

A demo Clean, controlled and rehearsed
A deployment Dust, noise, bad lighting and surprises

When it fails

A demo Cut the scene and try again
A deployment A line stops and someone gets a call

Who relies on it

A demo The team that built it
A deployment Workers, customers and a budget

In the field

Eight industries where AI has already left the lab. For each one, these are the kinds of jobs Kinetica watches for — work that happens every day, whether anyone is filming or not.

Manufacturing

Robots and vision systems working real shifts on the factory floor.

  • Palletizing
  • Defect detection
  • Machine tending

Logistics

Picking, sorting and routing in warehouses and fleets, day and night.

  • Order picking
  • Parcel sorting
  • Route optimization

Healthcare

Models that help with intake, imaging and paperwork in working hospitals.

  • Scan triage
  • Clinical notes
  • Scheduling

Energy

AI that forecasts demand and keeps power grids balanced.

  • Load forecasting
  • Line inspection
  • Battery dispatch

Agriculture

Machines that see every plant in a field and act on what they find.

  • Weed targeting
  • Yield estimates
  • Autonomous tractors

Retail

Systems that keep shelves stocked and stores running smoothly.

  • Shelf scanning
  • Demand planning
  • Self-checkout

Construction

Tools that measure, survey and check a site faster than a crew can by hand.

  • Drone surveys
  • Progress tracking
  • Safety monitoring

Customer service

Voice and chat agents that handle real calls around the clock.

  • Call answering
  • Booking
  • Support tickets

Field reports

Six examples of AI that has already left the lab and gone to work. Each one is a real, publicly known deployment — not a concept video.

Transportation · United States Scale

Driverless rides

Waymo

Fully driverless robotaxis carry paying passengers through several US cities, with nobody behind the wheel.

Why it matters: It proves AI can make real-time decisions in public, messy, unpredictable streets.

Healthcare logistics · Africa and US Scale

Medical drone delivery

Zipline

Autonomous drones fly blood and medical supplies to clinics in Rwanda and Ghana, and now deliver to homes in the US.

Why it matters: AI is solving a problem roads couldn’t, where minutes can save lives.

Agriculture · North America Production

See & Spray

John Deere

Cameras on a sprayer boom tell crops from weeds and spray only the weeds, cutting the amount of herbicide used.

Why it matters: Computer vision on a tractor turns a field into thousands of tiny decisions per second.

Logistics · Global Infrastructure

Warehouse robotics

Amazon

Fleets of mobile robots move shelves and totes across fulfillment centers, working alongside human pickers.

Why it matters: It’s so routine now that most customers never think about it — the final rung of the ladder.

Manufacturing · United States Scale

Robots as a service

Formic

Factories pay for robot work by the hour instead of buying the robot, making automation possible for smaller manufacturers.

Why it matters: A new business model can matter as much as a new model.

Retail · United Kingdom and beyond Scale

Robotic grocery picking

Ocado

Swarms of robots move across a giant grid, picking grocery orders inside automated warehouses.

Why it matters: Hundreds of robots coordinating at once shows AI working as a system, not a single machine.

Featured companies are examples only. None of them are affiliated with or endorse Kinetica.

The deployment ladder

Every AI system climbs the same five steps on its way into the real world. Most never make it past the second. Kinetica tracks where each one stands.

  1. 1 Demo It works once, under perfect conditions, in front of a camera.
  2. 2 Pilot A real customer tries it on one site, with the builders standing by.
  3. 3 Production It runs on a schedule and the customer depends on it to get work done.
  4. 4 Scale It spreads across many sites, shifts and customers without hand-holding.
  5. 5 Infrastructure It becomes so ordinary that nobody notices it anymore. That’s the goal.

What counts as deployed

Before Kinetica calls something real, it has to pass four checks. No exceptions for big names or good marketing.

A real customer

Someone outside the company uses it and pays for the result.

On a schedule

It runs regular shifts or hours, not only when someone presses play.

Without the builders

It works when the team that made it isn’t in the room.

Real work changed

It replaced or extended an actual workflow, not a side experiment.

Kinetica Tracker

An open list of AI that has passed all four tests. Every entry is placed on the deployment ladder. Filter by industry and region to see where AI is working.

13Deployments tracked
8Industries covered
8At Scale or beyond
5In Production
Waymo — Driverless rides Transportation · North America Fully driverless robotaxis carry paying passengers through several US cities.
Scale Passed all four tests
Zipline — Medical drone delivery Healthcare · Africa Autonomous drones fly blood and medical supplies to clinics in Rwanda and Ghana.
Scale Passed all four tests
John Deere — See & Spray Agriculture · North America Cameras tell crops from weeds so the sprayer hits only the weeds.
Production Passed all four tests
Amazon — Warehouse robotics Logistics · Multiple regions Mobile robots move shelves and totes across fulfillment centers.
Infrastructure Passed all four tests
Formic — Robots as a service Manufacturing · North America Factories pay for robot work by the hour instead of buying the robot.
Scale Passed all four tests
Ocado — Robotic grocery picking Retail · Europe Robot swarms on a grid pick grocery orders in automated warehouses.
Scale Passed all four tests
Starship Technologies — Sidewalk delivery robots Logistics · Multiple regions Small wheeled robots deliver food and groceries on campuses and city streets.
Scale Passed all four tests
Carbon Robotics — LaserWeeder Agriculture · North America Computer vision finds weeds and lasers remove them, with no herbicide.
Production Passed all four tests
Viz.ai — Stroke detection Healthcare · North America AI reviews brain scans and alerts stroke teams as soon as it spots a likely blockage.
Scale Passed all four tests
Skydio — Autonomous inspection drones Construction · North America Drones fly themselves around bridges, towers and structures to capture inspections.
Production Passed all four tests
Boston Dynamics — Spot inspection rounds Energy · Multiple regions A four-legged robot walks routine inspection routes at plants and industrial sites.
Production Passed all four tests
Google DeepMind — Data center cooling Energy · Multiple regions AI adjusts cooling systems in Google data centers to cut the energy they use.
Production Passed all four tests
Locus Robotics — Warehouse picking robots Logistics · Multiple regions Autonomous mobile robots guide workers to items and carry orders through warehouses.
Scale Passed all four tests

No deployments match this combination yet. Try another industry or region — or send us one on X.

Stages are Kinetica’s own assessment based on public information, and are updated as the field changes. Listed companies are not affiliated with Kinetica.

Why this is happening now

Robots and machine learning have existed for decades. What changed is that three forces finally lined up at the same time.

Models got good enough

Vision, language and planning models can now handle the messy, unpredictable conditions of real workplaces.

Hardware got cheap enough

Sensors, cameras, chips and robot arms cost a fraction of what they used to, so the math finally works for ordinary businesses.

The work needs doing

Many industries struggle to fill repetitive, physical and night-shift jobs. AI is stepping into the gaps people don’t want.

78%

of organizations reported using AI in at least one business function in 2024, up from 55% a year earlier. Source: Stanford HAI, AI Index Report 2025.

Intelligence keeps getting cheaper.

Every drop in the cost of running AI puts it to work somewhere new. A task that was too expensive to automate last year can become the obvious choice this year. The Abundance dashboard follows that curve.

280× cheaper to run GPT-3.5-level AI, November 2022 to October 2024
$0.07 per million tokens in October 2024, down from $20
30% yearly drop in the cost of AI hardware
40% yearly gain in AI hardware energy efficiency
Price of GPT-3.5-level AI, per million tokens
November 2022$20.00
October 2024$0.07

The same level of performance became more than 280 times cheaper in about two years. Source: Stanford HAI, AI Index Report 2025.

What’s pushing the cost down

Faster chips

Each new generation of hardware does more AI work for the same power and price.

Smaller models

Compact models now handle tasks that once needed the biggest systems available.

Open models

Freely available models let anyone build and deploy without asking permission.

AI on the device

More intelligence runs directly on robots, cameras and phones, with no data center required.

What we stand for

Proof over promises

If it isn’t running somewhere, it isn’t news yet.

Ordinary is the goal

The best AI is the kind that becomes boring and reliable.

Credit the builders

Behind every deployment are engineers and operators who made it work.

Respect the workers

We care about how AI changes jobs, not only what it automates.

Stay independent

No company pays to be featured. Every story stands on its own.

Open by default

What we learn, we share with the whole community.

Why Kinetica

The name comes from kinetic — energy in motion.

Kinetica started with a simple frustration. Every week brings a new AI announcement, a new benchmark and a new promise about the future. Meanwhile, some of the most impressive AI in the world is quietly sorting parcels, spraying weeds and driving passengers — and almost nobody talks about it.

We think the real story of AI is told in motion: machines and models that move goods, move people and move industries forward. Potential energy is a promise. Kinetic energy is work being done.

So Kinetica does one thing: it finds AI that is already doing the work, checks that it’s real, and shares it with a community that cares about results more than hype. It’s independent, community-driven and open to anyone who notices AI at work in their own corner of the world.

Questions

What is Kinetica?

Kinetica is a community project that follows artificial intelligence already working in the real world — in factories, warehouses, hospitals, farms and more — and shares those stories in one place.

Why focus on deployed AI instead of new models?

New models get plenty of attention already. What gets less attention is whether AI actually holds up in daily work. That’s where the real change happens, and where Kinetica looks.

How do you decide what counts as deployed?

Every system has to pass four checks: a real customer, a regular schedule, running without its builders in the room, and a real workflow changed. You can see where each one stands on the deployment ladder.

Is Kinetica affiliated with any company?

No. Kinetica is independent. The companies and industries we talk about are not partners or sponsors, and none of them pay to be featured.

Can I submit a deployment?

Yes. Share it with us on X. Field reports from the community are a core part of the project.

Is there a Kinetica token?

Not yet. Any official token details will be announced only through Kinetica’s official channels. Be careful of anyone claiming otherwise.

What can I find on Kinetica?

Field reports on AI at work, the Kinetica Tracker — an open list of real deployments you can filter by industry and region — and the Abundance dashboard, which follows the falling cost of AI using public data.

How to submit a field report

Anyone can add to the map. Here’s how a sighting becomes a Kinetica entry.

  1. 1 Spot it You see AI doing real work — in your job, your city, a news story or a company update.
  2. 2 Check the four tests Real customer, on a schedule, works without its builders, real work changed. If it passes, it’s a candidate.
  3. 3 Send it in Post it on X and tag Kinetica, with a link to the source.
  4. 4 Get it on the map We review the evidence, place it on the deployment ladder and add it to the Kinetica Tracker with credit to you.

Spotted AI at work? Tell us.

Kinetica is built by the people who notice. Share a deployment, follow the field reports and help map where AI is really working.

Follow Kinetica on X