Skip to main content

AI tools — description, social post, and pricing helpers

How to use the in-dashboard AI tools to write product descriptions, generate social captions, and suggest pricing.

Homegrown includes three AI tools inside your vendor dashboard to take repetitive marketing work off your plate: a product description generator, a social post generator, and a pricing helper. They're built to give you a starting draft you can edit — not a final answer.

Description generator

Stuck staring at the description field on a new product? The description generator takes your product name, a one-line note about it, and any tags, and returns a full description in your voice — typically 2-3 short paragraphs.

How to use it:

  1. Open a product in Products

  2. Click Generate description in the description field

  3. Add a sentence or two about what's special (ingredients, story, source)

  4. Click Generate, review, edit anything that doesn't sound like you, click Save

Social post generator

Generates a caption you can post to Instagram, Facebook, or any other social platform. It asks a couple of questions about what you're posting (a new product, a market this weekend, a seasonal limited edition) and returns 3-4 caption options.

Pricing helper

Takes your ingredient cost, packaging cost, time-per-unit, and your local market context and suggests a price range. It accounts for direct costs plus a target margin and compares against typical pricing for similar products. Use it as a sanity check, not a rule — your local context is always going to know better than an algorithm.

What the AI is and isn't doing

It's drafting. It's not deciding for you. Every output is a starting point that should sound more like you after a quick edit. AI-generated copy that goes out unedited reads like AI-generated copy — and shoppers can tell.

Privacy

The AI tools don't store your store data permanently. Inputs are sent to the model to generate the response, then discarded. Your descriptions, prices, and product info aren't used to train any models.

Related

Did this answer your question?