
Use **Video Review** when you want to inspect a recorded validation clip, watch it with overlays and playback controls, and add review marks at the exact moment you notice an issue.

This page is designed for quick customer-facing review work, especially when a team needs to confirm what happened in a recorded session.

## Open Video Review

1. Open the data source.
2. Click the **Videos** tab.
3. Open **Validation Videos**.
4. Click the video you want to review.

The page opens with the selected clip in the center, review marks on the left, and the rest of the session videos on the right.

![Video Review page with marks, playback controls, and the Session Videos gallery.](/docs/data-sources/video-review/01-video-review-overview.png)

## What you can do here

- play, pause, and scrub through the current clip
- switch between previous and next videos in the session
- review existing marks on the left
- add a mark at the current time in the clip
- keep nearby session videos visible while you work

## Review the clip

Use the center player when you want to inspect the current video closely.

- scrub through the timeline to jump to a moment
- pause on a frame before you add a mark
- use the toolbar controls for zoom or playback views that are available for that session
- use fullscreen when you need a larger view

## Use the session gallery

The **Session Videos** panel on the right keeps the current recording session within reach.

- click a thumbnail to open another video from the same session
- use the time label below each thumbnail to move quickly to the right clip
- keep the session visible while you compare nearby recordings

## Add a review mark

1. Pause the clip at the moment you want to mark.
2. Click **Mark**.
3. Choose the review reason that fits what you saw.
4. If you choose **Other**, add a short note before saving.

Saved marks appear in the **Marks** section with their timestamp, so anyone reviewing the clip later can jump back to the same moment.

![Video Review mark dialog with review choices.](/docs/data-sources/video-review/02-video-review-mark-dialog.png)

## Tips

- Pause before you add a mark so the saved timestamp matches the exact moment you want to highlight.
- Use the session gallery to compare nearby validation clips without returning to the main Videos tab.
- Review the existing marks first when you are checking work another teammate already started.

## Read next

- [Validation](/docs/ai-vision/validation) - Return to the validation session list for this data source.
- [Image Review](/docs/data-sources/image-review) - Review still QA session images when you need frame-by-frame image validation.
- [Remote Control](/docs/ai-vision/remote-control) - Collect new recordings from the live device.


Review validation videos, move through recorded sessions, and add review marks while you watch the clip.

Video Review


Use **AI Routing Assistant** when you want to build routing faster without writing every rule by hand.

It helps you create a first draft from a customer spec, then keep refining the same draft from the same conversation.

## Open AI Routing Assistant

1. Open a supported vision data source.
2. Click the **Routing** tab.
3. Review the **Lane Catalog** first so the lane names are ready.
4. Click the **AI Assistant** button in the top-right toolbar.

![Routing tab with lane catalog and assistant button](/docs/ai-assistance/routing-assistant/01-routing-tab.png)

## Start with the lane catalog

The assistant uses the lane names from **Lane Catalog** when it creates or updates routing.

Before you ask AI to help:

- add any lanes you know you will need
- remove old lane names you no longer want used
- keep the lane names clear and consistent

This makes the draft easier to review later.

## Work in one conversation

When you open the assistant, it stays next to your routing stages so you can review the draft while you chat.

Use the same conversation for both jobs:

- create the first routing draft
- ask for follow-up changes after the draft is ready

You do not need to switch to a different assistant view after the first draft is ready.

![AI Routing Assistant open on the Routing tab](/docs/ai-assistance/routing-assistant/02-ai-assistant-open.png)

## Create the first draft

Paste a customer spec or describe the routing logic you want in plain language.

Good requests usually mention:

- the lane names
- defect rules
- measurement or spec checks
- model selection logic
- business routing rules
- the fallback behavior

Examples:

- Route broken pallets to **scrap-lane** and keep good pallets in **good-lane**.
- If measurements are invalid, send the pallet to **measurement-invalid-lane**.
- Use model selection before spec validation.
- Keep business rules after spec validation.

After you send the request, your message appears in the conversation right away and the assistant starts preparing a draft.

![Assistant generating a routing draft from a prompt](/docs/ai-assistance/routing-assistant/03-generating-draft.png)

## Review the draft before applying it

When the draft is ready, use **Draft Review** to check what the assistant created.

You can:

- expand a stage to inspect the drafted rules
- review the lane, disposition, and conditions
- copy the draft JSON if you want to compare it elsewhere
- discard the draft and try again

If the draft looks good:

1. Click **Apply Draft** to load it into the routing editor first.
2. Click **Apply & Save** when you are ready to save it.

![Draft review with expandable stage changes](/docs/ai-assistance/routing-assistant/04-draft-review.png)

## Ask for changes after the draft is ready

Keep using the same conversation box under the draft review.

This works well for follow-up requests like:

- send red pallets to **repair-lane**
- remove the odd-size business rule
- tighten one spec validation rule
- add a new exception for one customer or pallet type

You do not need to paste the full routing policy again unless you want to start over.

## Start over when needed

Use **Start New Draft** when you want the assistant to stop working from the current routing and create a fresh draft instead.

This is useful when:

- the current routing has drifted too far from the original goal
- the customer policy changed a lot
- you want a clean draft based on a new set of instructions

## Keep the final review simple

AI can save a lot of time, but the final routing still deserves a quick manual check.

Before you save, review:

- lane names
- stage order
- rule conditions
- dispositions
- fallback behavior

This helps you catch anything that should be tighter, simpler, or more explicit.

## Read next

- [Routing](/docs/ai-vision/routing) - Full routing workflow, lane catalog, stage editing, and manual rule changes.


Open the Routing tab, describe the routing logic you want, and let AI draft or refine stage-based routing rules in one side conversation.

AI Routing Assistant


Training is how a camera learns to recognise something new — a defect, a part, a label. You show it
a dataset of images you have already marked up, and it produces a **model**: the file the camera
uses to make its own decisions.

The **Training** tab lists every training run this data source has done, and how each one went.

## Before you start

You need a dataset with labelled images, saved as a **version**. A version is a frozen snapshot of
the dataset — training always runs against one, so the run can be repeated later with exactly the
same images.

See [Datasets](/docs/ai-vision/datasets) for how to create one.

## Start a run

1. Open the data source and go to **Datasets**.
2. Open the versions list for the dataset you want to train on.
3. Choose **Run training** on the version.
4. Fill in the **Train a model** form (below) and start it.
5. Open the **Training** tab to watch it.

You can close the page once the run has started. It keeps going in the cloud, and anyone on your
team sees it in this tab.

## The "Train a model" form

Only one field is required — the name. Everything else already has a setting that works for most
jobs.

**What you are making**

| Field | What it means |
| --- | --- |
| Model name | What this model will be called in the Models tab. Give it something you will recognise in three months. |
| What this model looks for | Groups the model in the Models tab so it is easy to find. Optional, and you can set it later. |
| Dataset version to learn from | The images and labels the run learns from. |
| Camera view | Pallet cameras only — which side of the pallet this camera sees. |

**How it trains**

| Field | What it means |
| --- | --- |
| Starting point | Training rarely starts from nothing. Pick a stock model, or carry on from one of your own so it keeps what it already learned. |
| Image size | Bigger spots finer detail but trains and runs slower. 640 suits most cameras. Fixed when every image in the dataset is the same size. |
| Training passes | How many times it reads every image. 100 is a good start. |
| Stop if it stops improving | Ends the run early once this many passes bring no progress, so you are not paying for work that is not helping. |

**Also build a file the camera can run**

Training produces a model the cloud can read. Cameras need a smaller, converted copy, and this
builds it at the same time. Leave it on unless you only want the training result.

If you see a warning that the chosen format does not run on the camera app yet, the file will still
be built and stored — you just will not be able to send it to a device until the app supports it.

**Advanced**

Three collapsed sections — how it learns, which images it uses, and which cloud machines run the
job. Nothing in there needs changing for a normal run.

Inside them you can also override the **image variations** the run generates — small changes to
brightness, angle and framing that help a model cope with a real production line. The defaults are
tuned already; change them only when you know the variation you want to add.

## Follow a run

Each row in the Training tab shows the model being made, what it started from, how far it has got
and how long it has taken.

Click a row to open the full detail:

- **Progress** — which pass it is on, and a rough estimate of the time left.
- **Charts and epoch log** — how the model improved pass by pass, including a per-class view so you
  can watch one defect catch up (or fall behind) while the run is still going. Switch between the
  two scoring styles to see whether a class is merely being found or being placed accurately.
- **Image variations** — the variations this run generated from your images, listed so you can see
  what the model was trained against.
- **Dataset** — how many images went into training, validation and testing, plus a warning if the
  same image appears in more than one of them (which makes the scores look better than they are).
- **Final scores** — per class, how often the model was right when it fired, and how much it found.
- **Dataset sources** — if the dataset was merged from several others, which ones and in what split.
- **Restore points** — snapshots taken during the run. You can turn any of them into a model.

### "Stopped early" is normal

Most healthy runs end this way. It means the model stopped improving, so the run finished instead of
burning through the remaining passes. The result is still the best version it found.

## When it finishes

The new model appears in the [Models](/docs/ai-vision/models) tab. From there you can send it to a
device, or fine-tune from it to make the next version.

## Read next

- [Datasets](/docs/ai-vision/datasets)
- [Models](/docs/ai-vision/models)
- [Validation](/docs/ai-vision/validation)


Train a new model from a dataset, follow the run while it works, and understand the result.

Browse Guides

Video Review

Open Video Review

What you can do here

Review the clip

Use the session gallery

Add a review mark

Tips

Read next