> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bowerlabs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Protocol step videos — show me, don't tell me

> Attach a how-to video to a protocol, mark where each step happens, and ask Bird to play the right moment — by step number or by name, hands-free.

**Shipped 2026-06-12 · Protocols & Live Mode**

A protocol tells you *what* to do. Now it can show you *how*. Attach a
video of the protocol being performed — a training recording, a clip
from a previous run — and mark the moment each step begins, the way
YouTube chapters work. Markers save as you set them; there's nothing to
submit.

## Ask for the right moment, hands-free

Once a video is bound, all of this works by voice in a live session —
in any mode, gloves on, hands full:

* *"Show me the video for this protocol"*
* *"Show me this step"* — while running the protocol, plays the
  current step's clip
* *"Skip to step 3"* — jumps by number
* *"Show me the vortex step"* — jumps by name; a fragment is enough
* *"Play it from the start"*

Bird finds the clip, seeks to the marker, and plays it — on the web, on
mobile, and on AR glasses, where the video floats beside the protocol
card so the clip and your bench stay in view together.

## Setting it up takes a minute

Open a protocol and you'll see the new **Step video** section. Click
**Video recording** in the toolbar to attach your video, then play it
and hit **Set marker** as each step begins. During live runs, steps
with markers show a **Watch** button.

A couple of details:

* One video per protocol for now; H.264 MP4 plays most reliably
  across devices.
* Captures taken during protocol runs have moved off the protocol page
  — use the **⋮ → View captures** menu to see them in the Capture
  Stream, pre-filtered to that protocol.

[How to set up step videos →](/ai-and-search/protocol-step-videos)
