# Work the wishlist

> The assistant files what your store is missing, categorized for the team that can fix it - content gaps, feature requests, bug reports.

Source: CBX documentation, version 3.x (released). Canonical page: https://docs.configbox.at/docs/admin-guide/ai-assistant/work-the-wishlist. Last updated 2026-08-11.

---
> **Audience:** store owners, content managers, and whoever talks to the software vendor ·
> **Scope:** what wishlist items are, who each category is for, and how to act on them ·
> **Last reviewed:** 2026-08-11

Whenever a conversation exposes something the assistant could not do or answer — or something it
noticed was wrong — it files a **wishlist item**, invisibly to the visitor. Each item carries a
category that routes it to the people who can fix it, the assistant's description of the gap,
and, when a visitor request triggered it, that request in the visitor's own words.

Items appear in each conversation's timeline (collapsed to category and title — expand for the
full story) and are counted in the journal's **wish** and **unfulfilled** chips.

![A conversation that filed a wish: the chips, and the lessons review naming the gap](./img/conversation.jpeg)

The review above the timeline is where the gaps read best: it names the wishlist item that was
filed, and — as here — the ones the assistant *should* have filed and did not.

## The categories and who they are for

| Category | It means | Who acts on it |
|---|---|---|
| **content** | A visitor wanted product information your data does not cover — specifications, dimensions, delivery, stock, comparisons. | Content managers: add it to the product's detail panes or descriptions, and the assistant knows it from then on. |
| **feature-request** | A legitimate action the assistant had no way to do, or an ability that would have served the visitor better. | The software vendor — this is direct field evidence for what to build next. |
| **bug-report** | Something behaved wrongly: a price that does not add up, a selection that will not stick for no visible reason, contradictory data. | Developers / the vendor. |
| **translation** | Wording missing or wrong in the visitor's language. | Content managers / translators. |
| **ux-friction** | The visitor was confused or misled by how the configurator presents things. | Whoever owns the storefront. |
| **other** | Anything else worth passing on. | Triage. |

## How to work them

1. **Content items are the quick wins.** Each one names exactly what information was missing for
   which product. Put the answer into a detail pane and the gap is closed — the assistant reads
   panes as product knowledge.
2. **Unfulfilled items deserve priority** — the "unfulfilled" chip means a real visitor asked
   and left without. That is demand, measured for free.
3. **Feature requests and bug reports** are written to be forwarded: concrete descriptions of
   the missing ability or the observed misbehavior, with the conversation reference for context.
4. For volume, don't read item by item — run the
   [Insights report](https://docs.configbox.at/docs/admin-guide/ai-assistant/run-the-insights-report), which groups wishlist items across
   conversations, deduplicates common causes and ranks them by effect and effort.

The long-term idea: this diagnostic data is what your installation can eventually share with the
vendor, so content gaps, feature demand and field bugs surface across all stores — not just
yours.