About 8Kava

We build stores for the hours when nothing else is open.

About 8Kava in one line: we build AI-powered smart vending and micro-retail, meaning the Kava OS software and the connected hardware that let a business run an unattended store with very little overhead. Product, hardware lab and Network Operations Center work under one roof in Fort Collins, Colorado.

connected retail points under management
3,400+connected retail points under management
operators and location hosts on Kava OS
190+operators and location hosts on Kava OS
transactions in the trailing 12 months
11.8Mtransactions in the trailing 12 months
Miniature mountain hut in night snow with its door ajar, amber light spilling over stocked shelves and a small cooler, and a mint lantern above the door.
The ideathe best shelters are always open, quietly stocked by people you rarely meet, and built to keep working when no one is watching.
A1Why we exist

Hunger doesn't keep business hours

Food at work, at school and at home still depends on whether someone is standing behind a counter. Cafeterias close at 7 p.m. Night nurses, third-shift pickers, students and residents keep going. For decades, vending filled that gap with a spiral, a coin slot and a route driver guessing what had sold.

8Kava exists to make unattended retail worth trusting. That means checkout that charges the right item, real food held at a safe temperature, and software that tells an operator what each device needs before anyone starts the truck. Our tagline is the whole brief: retail that runs itself.

We sell software, Kava OS, as the core of what we do. We build hardware to put that software in more places, and we offer services for hosts who want an outcome rather than equipment.

A2Principles

Six rules we hold ourselves to

Open by default

Kava OS works with existing MDB machines and readers, takes your own payment processor at no markup, and exports your data. A customer who can leave is a customer we have to keep earning.

Honest numbers

Platform figures are aggregated across real deployments over the trailing 12 months, and case studies state their own window. 99.4% checkout accuracy is audited against restock counts, not estimated.

Privacy at the source

Recognition runs on the device and looks at products, not people. No images of faces are stored or sent. Analytics are anonymous and aggregated.

Built to be repaired

Camera, lock and compute boards are modular. A failed part is a swap, not a scrapped cabinet, and VaaS hardware is refurbished and redeployed at end of term.

Respect the route

We design for the people who restock at 5 a.m. If a feature doesn't save a driver a step, a stop or a phone call, it waits.

Name the trade-off

Some sites are wrong for a smart cooler. Some machines should be replaced, not retrofitted. We'd rather lose a deal than install a bad fit.

A3How we build

We wear things out on purpose.

Hardware engineers, machine-learning engineers, operations researchers, payments specialists and field technicians work in the same Fort Collins building. That matters more than it sounds. When a door hinge, a recognition model and a route plan are designed within earshot of each other, the trade-offs get settled at a workbench instead of in a support ticket.

New hardware spends its early life in the lab being opened, closed, warmed, chilled, disconnected and restocked badly on purpose, until the weak parts show themselves. The parts that fail become the next design.

  • Recognition fuses vision with per-shelf weight at ±2 g, so look-alike items don't become disputes.
  • Every device is built to keep selling through 72 hours offline.
  • Designed to UL-listed component standards, with R290 natural refrigerant in all refrigerated units.
Small smart cooler clamped in a brass test rig with a piston arm pulling its door, beside a museum-style tray of worn gaskets, hinges and lock bolts.
The ideawe wear our hardware out on purpose, so the one in your lobby has already survived its hardest years.
A4From problem to fleet

How an idea becomes a feature

  1. Hallway

    Start where the problem is

    We watch restocks, read service tickets and ride routes. Most features start as a complaint from someone holding a crate.

  2. Lab

    Break it before you do

    Prototypes face warm rooms, dropped connections, look-alike products and doors held open too long.

  3. Suggest

    Earn trust before autonomy

    New Kava Agent behaviors begin as suggestions a person approves. Auto comes later, and only inside guardrails the operator sets.

  4. Fleet

    Measure in the open

    Platform metrics are aggregated across deployments over the trailing 12 months. Case studies name their own window, such as the first 6 months.

B1Fort Collins

A lab and a NOC that never close either

Our Network Operations Center watches the connected fleet 24/7/365 from the same building as the lab. When a compressor trends warm or a door is held open, the alert reaches people who can read the telemetry and, if needed, call the engineer who designed the part.

In the field, 8Kava technicians serve the Colorado Front Range with same-day on-site response. Elsewhere in the continental US, vetted, 8Kava-trained field partners provide next-business-day service in served metros.

  • 4-hour remote response target for critical device-down issues.
  • 99.9% Kava OS uptime SLA, and 99.95% on Enterprise.
B2In their words

Night shift used to eat from a bag of chips at 3 a.m. Now they eat a real meal, and I get an audit-ready temperature log without anyone touching a clipboard.

Director of Food & Nutrition Services, Mountain West hospital network22 Kava Coolers and 4 Kava Lockers across 3 campuses. Revenue rose 46% over the prior vending program in the first 6 months.
Think about itSlot B4

Who in your organization eats last?

Every building has people who arrive after the food goes away: the night nurse, the third-shift picker, the student in the library at 1 a.m., the resident who gets home after the corner store closes. They're rarely the ones choosing the amenities. If you asked them what's missing, what would they say?

8Kava · always open
C1FAQ

About the company

Where is 8Kava based?

Our headquarters, hardware lab and Network Operations Center are at 2821 Remington Street, Suite 101, Fort Collins, Colorado. We deploy across the continental United States.

Is 8Kava a software company or a hardware company?

Both, in that order. Kava OS is the core recurring product, and it runs on MDB-compliant machines we didn't build. Our own hardware exists to put the software in places a legacy machine can't reach, like fresh-food coolers, lockers and walk-in micro-stores.

Who works at 8Kava?

Hardware engineers, machine-learning engineers, operations researchers, payments specialists and field technicians, all working from Fort Collins. If that mix sounds like your kind of team, see careers.

Do you run machines yourselves?

When a host wants us to, yes. Managed Operations covers stocking, merchandising, cleaning, service and reporting. Operators who run their own routes use the same Kava OS, with no one in the middle.

How do I reach you for press or partnerships?

Email hi@8kava.com. It's read by people who can answer, and it's also the right address for field service companies that want to work with us.

C3Next step

Come see the lab, or bring us your hardest site.

Tell us about the place where food runs out first. We'll show you what we'd build there, and what we wouldn't.