A technician standing at a machine with a fault rarely has time to search through a 300-page manual. The answer usually exists somewhere in the company, in a document, in a colleague’s experience or in photos from the last repair. Over the past months we have worked on Ari OS, RealWear’s operating system for AR headsets, to bring that knowledge to the worker hands-free, and to make sure what workers capture in the field gets back to the team.
Your documents, answerable by voice
With Ari Cloud Documents, teams can now upload their own files, from manuals and work instructions to spreadsheets, pictures and videos, even whole folders at once. Ari reads each file and prepares it for search. For example, we uploaded a free diving manual and asked Ari to find the section on helping a diver who has passed out. Ari gave the short version of the rescue steps, named the page and the section, and linked to the source.
A summary is only useful if you can check it, so the worker can follow up with “show me the exact place”. Ari opens the page, highlights the relevant passage in yellow and zooms in on request. The same works for video. Ari transcribes uploaded recordings, so a question like “where does the trainer explain how to send the reply?” opens the video at the exact second where that step begins.
Answers are grounded in the material a team chooses to upload, and companies decide what goes into Ari Cloud and what stays out. Search also works across languages, so a German manual can answer a question asked in English. Document answers are available today in Ari Chat on the web, and the same experience is coming to the headset next.
Capture once, find it again later
Photos and videos taken on the headset now show up in the worker’s Ari Cloud by themselves, with no cables and no manual transfers. An inspector can document a full round hands-free and find every picture ready at the desk afterwards. A technician can film a tricky repair and use the recording later to show a new colleague how it was done.
Captures are also easy to find again. A supervisor can search for “corroded valve” and get the right photos from last week’s inspection without scrolling through hundreds of files. Over time, these pictures and videos build up a visual record of the site that helps with shift handovers, audits and training.
Expert know-how as guided workflows
With Skill Workflows, an expert describes a procedure to Ari in plain language, and Ari turns it into a step-by-step workflow that appears as a visual flowchart in Ari Cloud. Changes happen in the same conversation, for example “add a safety check before step four”, and every edit is highlighted and easy to undo. Next, we will run these workflows on the headset, where Ari walks through the procedure with the worker and can still answer questions within each step.
Built around the worker
A few principles guide this work, and they are close to what the XR5.0 community cares about:
- Ari shows where its answers come from, and personal content stays private until its owner decides to share it.
- What workers capture in the field becomes searchable for the next shift, so they add to the company’s knowledge as well as use it.
- The same features run on the RealWear Navigator 520, the Navigator Z1 and the Arc 3.
How this connects to XR5.0
XR5.0 sets out to make extended reality human-centric, AI-driven and ready for Industry 5.0. Partners in the project have looked at knowledge management for remote maintenance and at conversational agents, and document answers and searchable captures bring that to a headset. Adaptive guidance and training, central to many XR5.0 pilots, is where Skill Workflows are heading. The project’s focus on trustworthy AI is why Ari cites its sources and why data stays under the control of the people who created it.
Talking with the other partners has helped us a lot. For every new feature we ask whether it makes the worker’s day easier, and the feedback from the consortium keeps that question honest. We will share more as XR5.0 moves into its final evaluation phase.