"How does this widget work?"
Ask it. Watch the answer draw.

Send a Widget Blueprint over from Unreal - a HUD you inherited, a menu from an asset pack, an Editor Utility Widget someone left behind - and ask the AI you already use how it fits together. The lesson draws on both halves at once: rings on the event graph, callouts pinned to the live preview, so the button you can see and the logic behind it are explained as one thing.

Widgets mode · One-click send from Unreal · Live preview · Editor Utility Widgets too · Free

A real lesson, on a real widget

WBP_HealthBar
Health
DamageHeal
Designer preview · WBP_HealthBar
On Clicked Btn_Heal
Set Percent Prg_Health · 1.0
Target
In Percent
On Clicked Btn_Damage
Set Percent Prg_Health · 0.25
Target
In Percent
Same tree, same look - so why is there no Tick anywhere in this widget?
Heal is an event, not a poll - this chain runs only on the click.
Set Percent writes the bar directly - no binding, no per-frame cost.
The preview is live - this exact tree is what deploys to Unreal.
Tree defines look - events define behaviour - nothing runs per frame.
Why there's no TickA real lesson, drawn on the graphclick ▶ to step through it
Why there's no Tick1 / 5

Step through it with the arrows - rings on the graph, callouts on the preview, exactly as it lands on your own widget.

From a widget you can't read to one you can

1
Send it over

One click in the Unreal editor (free Connector plugin) sends the Widget Blueprint to GraphDeck - the widget tree with layout and slot properties, plus the event graph that drives it, with a live preview assembled from the tree.

2
Ask how it fits together

Ask for an explanation of the whole widget, or a specific question: "What happens when this button is clicked?" "What drives the health bar?" The AI you connect reads the real tree and graph - names, bindings, and wiring - not a screenshot.

3
The lesson lands on both halves

Rings and notes on the event-graph nodes, callouts pinned to the live preview itself - so layout and logic are explained together. You page through at your own pace, and lessons can be saved and replayed.

4
Act on what you learned

When a lesson calls for it, the AI can sketch a suggested addition in chalk - nothing changes unless you accept it. Edit the tree and graph, then deploy the widget back into Unreal in one click. And when you want something new, the same AI can generate a whole widget from a description.

Use AI to understand Unreal rather than avoid understanding it

Widget Blueprints split the story in two: what you see lives in the designer tree, why it changes lives in the event graph, and the thread between them runs through bindings that are easy to lose track of. GraphDeck's Explain & Ask lessons - powered by the AI you connect, not an AI of GraphDeck's own - answer on both at once, anchored to the actual nodes and the actual widgets, so the explanation and the widget are never two separate things you have to reconcile.

Every mark is validated against the live graph before it draws. If the AI names a node that does not exist, the lesson is refused - your widget is always the ground truth.

The same explainer, in every graph system

Unreal's other node graphs get the same treatment - each drawn the way Unreal draws it, so the lesson lands on a graph you recognise.

Common questions

How do I understand a Widget Blueprint someone else made?

Send it over from Unreal - one click with the free Connector plugin brings the widget tree, layout, and event graph into GraphDeck together, with a live preview assembled from the tree. Then ask. The AI you connect explains it as a lesson drawn on both halves: rings on the event-graph nodes, callouts pinned to the preview itself. It works the same on a HUD you inherited, a menu from an asset pack, or a widget you generated.

Can AI explain a UMG widget?

Yes. GraphDeck has no AI of its own - you bring the one you already use: an assistant connected over MCP such as Claude or Cursor, any free AI chat via the copy-prompt workflow, or your own API key. GraphDeck turns its answer into an on-canvas lesson, and every mark is validated against the live graph before it draws - if the AI names a node that does not exist, the lesson is refused.

What does a widget lesson look like?

Widget lessons draw on both halves of the widget at once: rings and notes on the event-graph nodes, callouts pinned to the live preview itself - so the button you can see and the logic behind it are explained as one thing. You page through at your own pace, and lessons can be saved and replayed.

Does it work on Editor Utility Widgets?

Yes. Editor Utility Widgets are Widget Blueprints, and GraphDeck treats them the same way - bring one in, ask how the tool works, and the lesson lands on its tree and graph like any other widget. See Widgets mode for the full Editor Utility Widget story.

Can it change the widget too, or only explain it?

Understanding is half the job. When a lesson calls for it, the AI can sketch a suggested addition in chalk - nothing changes unless you accept it. Edit the tree and graph on the canvas, then deploy the widget back into Unreal in one click. And when you want something new, the same AI can generate a whole widget from a description.

Is the explainer free?

Yes. Explain and Ask ship in the free edition, and the copy-prompt workflow means no API key is required. Free keeps 5 saved graphs in each tool and includes AI builds (10 a day, 100 in total); buying GraphDeck is a one-time purchase that removes the caps.

Open your widgets in it today

GraphDeck Free

The free edition receives, edits, validates, and explains Widget Blueprints with no API key and no subscription. Pair it with the free Connector plugin for one-click send from Unreal.