Skip to main content

AI content

Write a prompt once Every product gets the same voice

Ready-made prompt templates for descriptions, titles, short descriptions, SEO meta and attributes, saved to your library and run against each product's own data.

  • Your, community and curated collections
  • Runs on each product's own data
  • Opens straight into Prompt Lab

Key features

  • Save what works

    Write a prompt, or save a community or curated one as your own copy, and it stays in the library, shared with your project by default.

  • Product data comes along

    Every run sends the AI each product's name, description and attributes as data, so one template in plain language covers thousands of SKUs.

  • Apply it in bulk

    Select products, pick a template, generate. The same prompt opens in Prompt Lab when you want to tune it.

Why a library

Good prompts get lost

A prompt that wrote a good description last quarter is only worth something if the next person can find it and run it again.

  • New products arrive with three fields

    A supplier file gives you a name, an EAN and a price. Applying the description and SEO templates in bulk turns that into sellable copy the same afternoon.

    Result: every import batch starts at the same quality, whoever loads it.

  • Everyone writes their own prompt

    Two people asking for a product description in their own words get two different tones. A saved template pins the tone, the length and the attributes it may use.

    Result: one approved wording per content type.

  • Campaign copy gets rewritten per product

    Write the seasonal promotion prompt once, apply it to the products in the campaign, and change the wording in one place the next time it runs.

    Result: a campaign edit touches one template, not every product in it.

How it works

From template to published copy

Three steps, and the last one runs on as many products as you select.

  1. 01

    Pick a collection

    ~1 min

    Open the library, search by title or description, and narrow the list by field, industry and purpose. Every prompt shows the field it writes.

    You have a template that already matches the job.

  2. 02

    Make it yours

    ~3 min

    Edit the wording, set tone, purpose and length, and add what your brand always or never says. Save it under your own name.

    Your version sits in Your Prompts and is shared with the rest of your project.

  3. 03

    Run it on the products you choose

    Select products in the grid, pick the saved prompt and generate. The run works in the background and writes the output to the field, product by product.

    Read a sample of the results before you export them.

What the templates cover

A starting prompt for every content field

Five fields and your own templates, each for content that has to be filled before a product can go live.

  • Product descriptions

    Benefit-first copy built from the attributes you already store, with length and structure fixed by the template.

  • SEO titles and meta

    Meta titles and descriptions that stay inside the character limits and use the keywords held on the product.

  • Short descriptions

    Listing teasers and summary blocks, short enough to work at the top of a product page.

  • Specifications

    Pull dimensions, materials and technical values out of raw supplier text into the attribute fields that expect them.

  • Product titles

    Titles built from brand, product type and key attributes, with a channel's length limit written into the template.

  • Your own templates

    Start blank, or save a curated prompt as a copy and change it. Your edits never touch the original.

Curated collections

Start from a maintained prompt

Curated collections group the prompts we keep up to date for common e-commerce jobs, so a new workspace never starts from an empty list.

A saved prompt becomes yours: it lands under your project defaults with its own version number, so a later update to the curated version cannot overwrite your edits.

AI Wizard

Describe the outcome, get a draft

Answer a few questions about the product type, the audience and what the text has to do, and the wizard writes a template you edit before saving.

  • Answer three questions

    Goal, audience and content type. No prompt-writing vocabulary needed.

  • Edit before you save

    The draft opens in the normal prompt editor, where you change the wording before it is saved.

  • Test it on a real product

    Send it to Prompt Lab, run it against one SKU, and save the version that held up.

The answers sit on the left and the drafted prompt on the right; nothing enters your library until you save it.

Product prompt templates

What a product prompt template is

A chat prompt is written for one answer. A product prompt template is written once and run on thousands of products, so it has to work without the product being in it.

In WISEPIM a prompt template is the instruction part of a generation job: what to write, for whom, in what shape and with what emphasis. It is stored with a set of settings that steer the output (the field it writes, tone, purpose, industry, point of view and length) and it is run against a selection of products.

The product itself is not pasted into the prompt. Every run sends the AI the product's name, description, all its attributes and any web research findings as separate, structured data, along with your Knowledge Library, your project keywords and your Quality Guard rules. That is why one template can cover a whole catalog: the instruction stays the same and the facts change per product.

This also means you can write a template in plain language. You do not need placeholders such as a product name variable to make the AI look at the product; it already does.

The settings saved with every prompt template
SettingOptionsWhat it changes
FieldDescription, short description, titles, SEO title and meta description, product attributes, FAQsWhich product field the output is written to, and which rules apply
ToneTwelve presets, or a custom tone you describeThe voice, from technical to playful
PurposeEighteen angles, such as SEO, conversion, technical detail or storytellingWhat the copy is trying to achieve
IndustryYour sectorVocabulary and which product details matter
Point of viewSecond person or third person"You get" versus "This product offers"
Target lengthAuto, about 80, 150, 300 or 500 words, or a custom countHow long descriptions come out, so one prompt fits several channels
ExtrasSpecification table, title variants, FAQ typesField-specific options for the job at hand
The settings saved with every prompt template

Free prompt templates

More than 100 curated templates to start from

Nobody should start from an empty box. WISEPIM ships a maintained collection of prompt templates for the fields every product needs.

The curated collection holds 104 prompt templates written by the WISEPIM team: 32 general ones and 72 written for 20 specific industries, from fashion and furniture to industrial supplies. Inside the app the collection is ranked for your project, so templates that match your industry come first.

You can browse the same collection without an account in our free prompt templates directory, grouped by the field each template writes. Signing up gives you the full text and lets you save a template to your own library and run it on your products.

Curated prompt templates by field
FieldTemplatesTypical use
Descriptions74Long product descriptions, general and per industry
Short descriptions9Listing teasers and summary blocks
SEO title and meta description9Search snippets that match the product page
Product attributes8Pulling structured values out of text and data
Titles4Product titles built from brand, type and key attributes
Counts of the curated collection in October 2026. Templates that are retired are hidden from the list, not deleted.

There are no curated translation templates on purpose. Translation runs through its own engine with a dictionary, translation memory and per-language rules; see AI product translation.

Writing a good template

The anatomy of a product prompt that holds up at scale

A prompt that works for one product often fails on the next one. These are the parts that make it hold across a catalog.

  • One job per template. A description template writes descriptions. Keep titles, meta descriptions and FAQs in their own templates so each one can be tuned and tested alone.
  • Name the reader. "Written for facility managers comparing replacement parts" changes the output more than any adjective about tone.
  • Say what to lead with. Tell the AI which facts matter most in your category, such as fit and fabric for clothing, or load rating and certification for tools.
  • Give a structure. An opening sentence, three benefit paragraphs and a closing line about care is easier to check across 5,000 products than free prose.
  • List what must never appear. Claims you cannot prove, competitor names, words your brand avoids. The AI follows a short, explicit list better than a general request to be careful.
  • Leave the length to the setting. Use the target length instead of writing "about 150 words" into the text, so you can reuse the template for a shorter channel.
  • Put channel limits in titles. Marketplace titles have hard limits; Amazon, for example, caps most titles at 75 characters from 27 July 2026. Write the limit into a title template for that channel.

Search engines judge the result, not the method. Google's guidance is that AI-written content is fine when it is accurate and useful, and that mass-produced pages with little value are not. A template that makes the AI use the product's real attributes is the practical way to stay on the right side of that line.

From template to catalog

Six steps from a curated template to finished copy

The route most teams take the first time, from picking a template to running it on a selection.

  1. 01

    Find a template

    Open the Prompt Library and search by title or description. Narrow the list by field, industry and purpose. Curated templates sit in their own tab.

  2. 02

    Save it as your own

    Save a curated template to Your Prompts. You get a copy you can edit; the original stays untouched and keeps being maintained.

  3. 03

    Make it fit your catalog

    Change the wording, set the tone, purpose, industry, point of view and target length, and add what your brand always or never says.

  4. 04

    Test it on real products

    Open the template in Prompt Lab, load a real product and compare the output with the current value. Try a thin product as well as a rich one.

  5. 05

    Make it the default

    Set it as the default for its field. The default is preselected in the enrichment dialog and used when WISEPIM fills gaps on its own, such as after a scraped import.

  6. 06

    Run it

    Run it from the library on products you search and pick, from the product grid on a selection or a whole filter, or as a step in an automation. Generation uses 1 AI credit per product.

AI help inside the library

Tools for writing and improving templates

Writing a template is a skill. The library has AI tools that do the first draft and the second opinion.

  • AI Prompt Wizard

    Answer questions about your goal, audience and field, and the wizard drafts a template you edit before saving.

  • Improve

    Ask AI to sharpen an existing template and see the changes as a diff before you accept them.

  • Variants

    Generate alternative versions of a template to test against each other in Prompt Lab.

  • Remix

    Combine the strengths of templates you already have into a new one.

  • Score and critique

    Get a score and concrete criticism of a template, such as missing audience or vague structure, before you run it on a catalog.

  • Auto-fill details

    Let AI suggest the title, description and settings for a template from its text, so the library stays searchable.

Defaults per field

One default template for each field

A project keeps one default template per field type, so the same wording runs no matter who starts the job.

  • Five defaults: description, titles, product attributes, SEO title and meta description, and short description.
  • Preselected everywhere: the default appears first in the enrichment dialog for its field, so a colleague who never opens the library still uses the approved wording.
  • Used by gap filling: when WISEPIM fills empty fields on its own, for example after an import by scraping, it uses the defaults.
  • Changed in one place: set a default from the template card in the library or from the enrich prompt settings; both edit the same setting.

Sharing and versions

How templates are shared and kept

Templates belong to a project. What you can and cannot do with them today:

  • Shared with your project by default. A saved template appears under Team Prompts for the other members, who can run or duplicate it; only you edit it. Make it public and it also shows in the Community tab, or switch sharing off to keep it to yourself.
  • Duplicate before a big change. Duplicating keeps the working version next to the experiment.
  • Version numbers move with edits. Saving a change stores the previous configuration and raises the version shown on the card.
  • Set as default to make a template the standard for its field across the project.
  • Not available today: tags, folders, favourites, importing or exporting templates, a visible version history with restore, and an approval step before a template can be used.

What the AI reads

Your template is one of several inputs

Output quality depends on more than the wording of the template. Each run combines these inputs:

  • The product's data

    Name, current description and every attribute value, sent as structured data. A product with ten attributes gets better copy than one with two.

  • Web research findings

    Specs and facts gathered by web research, when research has run for the product.

  • Your Knowledge Library

    Brand voice, audience and product context from the Knowledge Library, read on every run.

  • Project keywords

    The keywords you store on the project, so SEO-minded templates use the terms you want to rank for.

  • Quality Guard rules

    Your content rules, such as length limits or forbidden words, so the output is written to pass them.

  • The project language

    Output is written in the project's language. Other languages come from the translation step.

Choosing the right tool

When a prompt template is not the answer

Some jobs look like prompting but are better solved elsewhere in WISEPIM.

  • The same sentence on every product

    A shipping line or a warranty sentence should not be generated. Write it once with Content Logic and it fills in per product.

  • Facts the product does not have

    A template cannot invent a weight or a material. Fill the gaps with EAN and web research first, then generate.

  • Another language

    Write once in your main language and translate with the dictionary and translation memory, rather than prompting per language.

Sources

Search and marketplace facts on this page

Google's guidance and the Amazon title limit come from the sources below, checked in October 2026.

FAQs

Prompt Library questions, answered

Collections, sharing, the AI Wizard, and how prompts move between the library and Prompt Lab.

Still have questions?

Can't find the answer you're looking for? Please get in touch with our team.

Contact Support

Your Prompts holds the prompts you saved. Prompts your colleagues share with the project appear under Team Prompts, and prompts made public in the project appear under Community Prompts. Curated Prompts are written and maintained by the WISEPIM team for common e-commerce jobs. Prompts you do not own are read-only until you save a copy into Your Prompts.

Use Create Prompt, give it a name and a short description, and write the instruction in plain language: every run already sends the AI the product's data, so there are no placeholders to fill. You can also open a community or curated prompt and save it as your own starting point. There is no approval step; a saved prompt is shared with your project right away unless you switch sharing off.

Yes, and it is the default. A prompt you save is shared with the other members of your project, who find it under Team Prompts. Make it public and it also appears under Community Prompts for everyone in the project. Only the owner edits a prompt; colleagues can use it or duplicate it into their own prompts. Switch sharing off on a prompt to keep it to yourself.

Search matches the prompt title and description, and you can narrow each tab by field, industry and purpose. There are no tags or folders, so a clear title and description are what keep a library findable.

Yes. Prompt Lab has a load button that pulls any saved prompt into the editor, where you can run it against a real product and compare the output with the current field. Saving from Prompt Lab can update the prompt you loaded or store it as a new one, so the tested version is the one your team runs.

The AI Wizard asks about your goal, your audience and the content type, then drafts a prompt template from those answers. It writes a starting point, not a finished prompt: the draft opens in the editor so you can change the wording and the settings before saving. It does not generate product content itself.

Browsing them is. The curated collection is published on our prompt templates page, with no account needed. Inside WISEPIM you save a template to your library and run it on your products; generating copy uses 1 AI credit per product, and the free plan includes 50 credits a month.

No. Every run sends the product's name, description, attributes and research findings to the AI as structured data next to your template, so the AI already knows which product it is writing about. Write the template as instructions.

Yes. Run it from the library on products you pick, from the product grid on a selection or a whole filter, or as a step in an automation for descriptions, short descriptions, titles, SEO meta and attributes.

Descriptions, short descriptions, titles, SEO titles and meta descriptions, product attributes and FAQs. Translation has its own engine and settings, so there are no translation templates.

From two places: the tone and purpose saved with the template, and the Knowledge Library, which is read on every run. Put the voice in the Knowledge Library once instead of repeating it in every template.

The library stores, organises and runs templates. The lab is where you test one template on one real product and compare the output with the current value before you run it on many.

WISEPIM picks the model for each job and keeps it up to date, so there is no model setting on a template. The settings you control are the field, tone, purpose, industry, point of view and length.

Next step

Share your best prompt with the team

Start from the curated collections, save the ones that fit your catalog, and run them on your own products.

4.7