PIING
Reading Piing: IntelligenceCustom fields and prompts
Piing: Intelligence
- How Intelligence works
- The fields it writes
- Profile refreshes and legacy units
- Preview your profiles
- Custom fields and prompts
- The storefront pixel
- Run and schedule analyses
- Where your data goes
In this guide
- Custom fields and prompts
- Output rules
- Why this matters
- Next steps
Search or browse documentation
Piing: Intelligence 5 / 8
Custom fields and prompts
On Scale plans and above, add your own AI prompts and rules to write custom customer fields, and use output rules to control what every field writes.
Piing: Intelligence ships with 22 ready-made fields. On Scale and above you can add custom fields and prompts, and on every plan output rules can control what reaches the Shopify customer record.
Custom fields and prompts
Custom fields are available on Scale plans and above. A custom field lets you analyse any aspect of your customer data that matters to your business, using your own wording.
You define a custom field by writing a prompt that tells the model what to look for and what values to return. The custom prompt runs alongside the core AI fields on every analysis run, and the result is written to its own Shopify metafield like any other field.
To create one:
- In Piing: Intelligence, open Custom Prompts.
- Add a prompt describing the field you want, the values it may return, and when to return null.
- Optionally start from one of the suggested prompts grouped by vertical, such as colour preference or fit preference for fashion, then edit it to fit your store.
- Save it. The field is created and runs on the next analysis.
Note
On new profile-refresh plans, custom fields are included in the customer refresh and do not add a separate refresh charge. Existing field-unit agreements continue to count each enabled field under their approved terms. See Profile refreshes and legacy units.
Output rules
Output rules are available on every plan and control what actually lands on the customer record. The AI proposes a value; your rules decide whether it is written, how it is worded, and how confident it must be. You manage them on the Rules page. There are four kinds:
- Block a value. Stop a specific value from ever being written, for example a value that does not fit your taxonomy.
- Map to your taxonomy. Rewrite an inferred value into your own wording on the way out, for example changing "Casual" to "Relaxed".
- Override the AI. Pin a value for a field so the model's output is ignored for that record.
- Confidence floor. Set the minimum confidence a value must reach before it is written. Anything below the floor is held back rather than saved as a weak signal.
Rules take effect on the next analysis run.
Why this matters
Used together, custom prompts and output rules let you build exactly the profile schema your brand needs: your own fields, your own wording, and a quality bar that keeps weak inferences out of downstream segments and campaigns.
Next steps
Related guides
Was this guide helpful?
Next guide in Piing: Intelligence
The storefront pixel
A lightweight theme app embed that captures browsing signals on your Shopify storefront and feeds them into richer customer profiles, with consent respected.
Still need a hand? Contact us