--- title: "Productify.ai Integration" description: "Productify.ai generates ecommerce product content from a name and an image. Agents fill in missing marketing copy, headlines, features and SEO keywords across a catalog, digitize product images into tables and back-of-pack data, and translate listings into 47 languages." url: https://flowrunner.ai/integrations/productify-ai date_modified: 2026-09-04T19:38:13-07:00 --- # Productify.ai [AI](https://flowrunner.ai/integrations/category/ai-llms) Productify.ai generates ecommerce product content from a name and an image. Agents fill in missing marketing copy, headlines, features and SEO keywords across a catalog, digitize product images into tables and back-of-pack data, and translate listings into 47 languages. 16 actions · API key · available [Productify.ai website](https://productify.ai/) · [Platform Documentation](https://productify.ai/) · Capability data verified 2026-08-26 1. A nightly sweep pulls catalog rows whose descriptions are empty or supplier boilerplate 2. Agent collects each product's name, image URL and category hint 3. Generate Product Content runs one representative product and returns the copy in the same call 4. Get Operation Costs and Get Account Balance price the full run against the remaining credits 5. The sample copy and the priced job go to the catalog owner as one message 6. The owner confirms the output selection and the spend, and only then does the batch get queued ## What This Integration Enables Productify.ai is built for catalogs rather than for documents. It takes a product name and an image and returns the specific outputs an ecommerce listing needs, each one switched on separately: the marketing description, the short description, the headline, the features and benefits, SEO keywords, and a taxonomy classification against the Google product taxonomy or GS1. It reads product photographs too, pulling table data, OCR text, nutritional panels and full back of pack information out of images it fetches from URLs the workflow supplies. FlowRunner agents run any of it one item at a time, with the result in the same call, or queue it as an asynchronous batch and poll for completion. The design decision that shapes every workflow here is that Productify.ai bills in credits per selected output, and it publishes its prices on a route that answers without a credential. That means an agent can cost a job honestly before it commits to one. It also means the platform has no cancel operation: once a batch is queued, the credits are spent. Pricing the run and confirming the selection is not bureaucracy, it is the only place a person can stand. ### Without FlowRunner **Copy written last**: Product pages ship with the supplier's paragraph because nobody had time to rewrite it **Spec sheets retyped**: Nutritional panels and back of pack data are keyed in from photographs by hand **Cost discovered afterward**: A bulk content run is priced by looking at the invoice next month ### With FlowRunner **Copy generated on intake**: New products arrive with a description, a headline, features and SEO keywords already drafted **Panels read from the image**: The label photograph becomes structured data without anybody transcribing it **Cost priced before the call**: The job is costed against the live balance and confirmed before a single credit is spent ## Use Case Scenarios ### Filling the gaps in an existing catalog The agent pulls products with thin or missing descriptions from [WooCommerce](https://flowrunner.ai/integrations/woocommerce) or [Shopify](https://flowrunner.ai/integrations/shopify), or from a supplier sheet in [Google Sheets](https://flowrunner.ai/integrations/google-sheets). It queues them with Create Ecommerce Batch, selecting only the outputs that are actually missing, because every selected output is charged separately. It then polls Get Batch Status until the status reaches Processed and reads the copy back with Get Ecommerce Batch Results, keyed by the product identifiers it supplied. Products that failed inside an otherwise successful batch are read back too, because a failure flag on the batch does not mean nothing succeeded. ### Turning supplier photographs into structured product data A supplier sends a folder of label photographs. The agent lists the image URLs, queues them with Create Digitisation Batch, and selects table extraction and back of pack data. Once the batch finishes, Get Digitisation Batch Results returns one structured record per image, keyed by the identifier the agent supplied so each result maps back to the product it came from. The records land in [Airtable](https://flowrunner.ai/integrations/airtable) as draft fields on the product row, ready for the merchandiser to check against the physical pack. ### Localizing a catalog without re-authoring it Before any translation runs, the agent calls Get Workspace Details, because translation is a workspace feature and a call that assumes it is enabled is refused at run time rather than at configuration time. It then reads List Supported Languages for the numeric ids, which are deliberately not derivable from a range, and queues the copy with Create Text Batch. Get Text Batch Results returns the translated strings, which are written back to the storefront's locale fields rather than overwriting the source language. ## Human-in-Loop Highlight The one way door on this connector is the batch create, and the reason is not that generated copy is dangerous. It is that Create Ecommerce Batch, Create Digitisation Batch and Create Text Batch each spend real credits on the whole list in one call, and there is no cancel operation anywhere in the connector's surface to take it back. A batch too large for the balance is refused in the response body rather than by an error status, so a workflow that only checks status codes can read a refusal as a success and move on. So the agent prices the job first. Get Operation Costs returns the per operation prices and the translation surcharge, Get Account Balance returns the remaining credits, and Generate Product Content runs a single representative product so the owner can read what the copy actually sounds like. Then the agent asks: "Four hundred and twelve products, description and headline and SEO keywords selected, translation off. Here is one product's output and the credits this will consume against the current balance. Queue it?" The person approves the selection set and the spend together, because on this platform they are the same decision. Agent processes routinely Detects exception requiring judgment Clear match Continues automatically Ambiguous Routes to human via preferred channel Human decides Agent resumes with decision ## Agent Capabilities 16 actions ### Single Operations 3 - **Generate Product Content** Generates ecommerce copy for one product and returns it in the same call, covering the marketing description, short description, headline, features and benefits, SEO keywords and taxonomy classification. Each output is selected and charged separately, which makes this the right call for a sample before a batch. - **Extract Image Content** Reads a single product image and returns the result in the same call, with table extraction, OCR text, nutritional panel data and full back of pack data selectable independently. Images are fetched from the URLs the workflow supplies, so no file upload is involved. - **Transform Text** Summarizes, rephrases or translates one piece of text and returns the result in the same call. Summarize and rephrase each return three options; translate returns one string. The input is capped, so longer copy is split first. ### Batch Operations 4 - **Create Ecommerce Batch** Queues a list of products for content generation and returns a batch id immediately, together with the credit cost and the balance left afterward. This is the operation this page's human gate protects. - **Create Digitisation Batch** Queues a list of image inputs for extraction, with the four extraction types selectable per batch. Returns a batch id immediately and processes asynchronously. - **Create Text Batch** Queues a list of text inputs for summarizing, rephrasing or translation, returning a batch id immediately. - **Get Batch Status** Reads a batch's numeric status, how many items have finished, and whether any failed. Productify.ai publishes no webhook subscription, so this is how a flow waits for asynchronous work. A failure flag means some items failed while others may have succeeded, so results are read either way. ### Batch Results 3 - **Get Ecommerce Batch Results** Reads a page of generated content from a completed ecommerce batch, one result per product keyed by the identifier the workflow supplied. - **Get Digitisation Batch Results** Reads a page of extracted content from a completed image batch, carrying whichever of the table, OCR, nutritional and back of pack extractions were selected. - **Get Text Batch Results** Reads a page of transformed text from a completed text batch, carrying the summarize, rephrase and translate outputs that were selected. ### Account 3 - **Get Account Balance** Reads the workspace credit balance with a per operation cost breakdown and an estimate of how many of each operation the balance affords. Doubles as the connection test and as the guard before a large batch. - **Get Workspace Details** Reads the workspace the API key belongs to, including its default language, unit of measure and whether translation is available. Worth checking before enabling translation, which is otherwise refused at call time. - **Get Operation Costs** Reads the public per operation credit prices grouped by batch type, plus the extra charge translation adds. This route answers without a credential, so a flow can price a job before an API key is even configured. ### Reference 3 - **List Supported Languages** Lists the languages available for translation, each with the numeric id the API expects. The ids are not contiguous, so they are read from here rather than derived. - **List Supported Categories** Lists the retail product categories that steer content generation, each with its numeric id. The ids skip values, so they are read rather than calculated. - **List Supported Taxonomies** Lists the taxonomies product classification can run against, which are the Google product taxonomy and GS1. The choice decides the shape of the category path returned on a generate result. ## Frequently Asked Questions ### What can FlowRunner do with Productify.ai? FlowRunner agents can run Generate Product Content, Extract Image Content, and Transform Text in Productify.ai, plus 13 more actions. ### Does connecting Productify.ai to FlowRunner require OAuth? No. Productify.ai connects to FlowRunner with an API key, no OAuth flow required. ### Can Productify.ai trigger a FlowRunner workflow automatically? Productify.ai doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger. **Work at Productify.ai?** This integration exposes Productify.ai to AI agents on every FlowRunner plan, including through MCP, at no cost to you. [See what FlowRunner offers integration partners](https://flowrunner.ai/integrations/partners), including how to keep this page current. --- Markdown version of https://flowrunner.ai/integrations/productify-ai. Site index: https://flowrunner.ai/llms.txt