Enterprise Adoption2 min read

Frontline Headsets Used to Give Orders. Now Workers Can Ask

By , Senior AI ConsultantPublished

Warehouse voice systems have given workers orders since the early 1990s; AI lets workers ask open questions and get answers from company policy, so know-how moves from supervisors to the floor.

AI that listens and talks is reaching workers who never had a keyboard: crews on power lines, truck drivers, factory staff, shop assistants. Samsara, which sells software for fleets and field crews, announced in October that on its platform voice agents have overtaken chat in use. Nobody can type with gloves on, and that is the whole reason.

Voice at work is old, but it could only take orders

The old warehouse systems worked because picking is predictable. Every answer is a number, a location or a yes, so a small computer only has to listen for a handful of fixed phrases. A question like "the label is torn, what do I do?" had no phrase to match, so it went to a person.

Open conversation needs two things the old systems did not have. One is speech recognition that copes with noise, accents and ordinary sentences. The other is a model that can find the answer in the company's own documents and act on it, for example by logging a fault. Both have improved a great deal in the last few years, which is why companies can now offer workers voice access in their own language, often on the phones they already carry.

What it looks like at a gate

On Samsara's platform, a truck that crosses a virtual boundary around a site the driver has never visited can trigger a phone call from an agent. The agent says which entrance to use, where to park, the yard speed limit and the safety gear required, and the driver can ask follow-up questions. When the truck leaves, the agent records what happened during the visit. In retail, the electronics chain MediaMarktSaturn has given sales staff voice assistants that look up product and stock details while they stand with a customer.

Now picture a warehouse supervisor in peak season with twenty people, half of them temporary. I would expect that supervisor to field about thirty how-do-I questions in a shift, at around two minutes each. That is an hour of the supervisor's day spent repeating answers that already exist in a binder nobody opens. A headset that answers them gives the hour back, and the new hire stops waiting or guessing.

Software built for desks left most workers out

Most of the world's workers do not work at a desk, yet company intranets, scheduling tools and policy portals were built for one. Questions like who covers Thursday, or what the rule is for a damaged pallet, went to a supervisor or a notice board. A voice assistant on the worker's own phone can answer them in the worker's own language. That matters most to a night-shift employee who reads English as a second language, and it is a bigger change for that person than for any office worker.

The first thing it shows is what nobody wrote down

An assistant that answers from the company's own policies is only as good as those policies. In the first weeks it will show every question workers have that the written rules never answered, or answered in two different ways. Managers rarely see that list, because today those questions disappear into hallway conversations.

A company that reads those questions every week and fixes the rules behind them gets better answers each month, and its new hires become useful sooner.

Share this

STAY INFORMED

Get AI intelligence like this delivered to your inbox.

Free forever · Unsubscribe anytime


You May Also Find Valuable