Freepik
AILet AI agents source and generate visuals through the Freepik API. Agents search Freepik's stock library of photos, vectors, and icons, pull download and license details for a chosen resource, and generate images from text prompts with Text To Image or high resolution Freepik Mystic.
What This Integration Enables
Freepik is one API where "find it" and "make it" are the same decision. Most visual pipelines treat stock search and AI generation as separate worlds with separate tools and separate owners; Freepik puts a stock library of photos, vectors, and icons next to two generation models behind one key, which means an agent can try the cheap answer first and escalate to a render only when the library comes up short. That decision logic is exactly what FlowRunner agents are built to run. Agents search resources and icons by keyword, pull full license and download details for a chosen asset, generate quick candidates synchronously with Text To Image, and run premium photorealistic renders through the asynchronous Mystic pipeline, polling the task until the URLs land. - Stock first, generation second: the agent only spends render credits when the library has no usable match - License metadata rides along with every sourced asset, so approval decisions include the legal facts - The Mystic polling loop is handled inside the flow, so downstream steps simply receive finished URLs - Every output routes through a [human-in-the-loop](/concepts/human-in-the-loop) review before it reaches a public surface
Without FlowRunner
With FlowRunner
Use Case Scenarios
Hero images that keep pace with the publishing calendar
A content team publishes through [WordPress](/integrations/wordpress). When a draft is marked ready, the agent calls Generate Mystic Image with a prompt assembled from the article's title and summary, then polls Get Mystic Task until the render completes. The finished image posts to [Slack](/integrations/slack) alongside the draft link. The editor approves or asks for a revised prompt. On approval, the flow uploads the image to WordPress and attaches it to the post. Writers stop sourcing their own imagery, and the editor still sees every visual before it goes live.
Product listings that never ship without imagery
New products land in [Shopify](/integrations/shopify) from a supplier feed, some without photography. The agent calls Text To Image for each bare listing, producing candidate visuals sized for the storefront grid, and logs the prompt and output in [Google Sheets](/integrations/google-sheets) so merchandising has a full record. The merchandiser reviews the batch once a day, approves the usable candidates, and flags products that need a real photo shoot. Nothing generated reaches the storefront without that pass.
Licensed assets for decks and docs, with the paper trail intact
An internal request flow collects asset needs from sales and marketing in [Airtable](/integrations/airtable). The agent runs Search Icons for interface glyphs and Search Resources for photos and PSDs, calls Get Resource on the shortlist to capture license type and author, and posts the preview set with license lines to the requester. The chosen asset's id, license, and download details are written back to the request record. When a question about usage rights arrives later, the answer is a lookup, not an archaeology project.
Human-in-Loop Highlight
Generate Mystic Image is the spend on this page that cannot be recalled: each render draws against a metered API balance, and the images it produces are usually destined for the most public surfaces a company has, hero banners, product pages, campaign creative. So the agent never runs an open ended generate and publish loop. It renders once, posts the result with the prompt that produced it, and waits: "Render complete for 'Q3 logistics report hero'. Style: photorealistic, 4k. Publish, or revise the prompt?" A designer either approves or edits the prompt, and only then does the agent spend again or push the image downstream. Machine speed for the rendering, human judgment for what the brand actually shows the world.
Agent Capabilities
6 actionsStock Content
3- Search Resources Searches Freepik's stock library (photos, vectors, and PSDs) by keyword. Supports pagination and per-page limits, and can filter results by content type. Returns matching resources with their id, title, thumbnail preview URL, and license and premium metadata.
- Get Resource Retrieves full details for a single stock resource by its numeric id, including title, author, available formats, dimensions, license type, and download information. The step that turns a search hit into an asset the team can actually clear.
- Search Icons Searches Freepik's icon library by keyword, with pagination and filtering by icon shape. Returns matching icons with id, name, style, and thumbnail preview. The fast path to consistent glyphs for decks, docs, and product surfaces.
AI Image Generation
3- Text To Image Generates images synchronously from a text prompt using Freepik's classic model, returning the images directly in the response. Choose how many images to produce, the output size and aspect ratio, and an optional visual style. The quick candidate generator for catalog and email visuals.
- Generate Mystic Image Starts a high resolution generation task with Freepik Mystic, the premium photorealistic model, at resolutions up to 4k with a choice of aspect ratio and rendering model. Asynchronous by design: it returns a task_id to poll, or calls a webhook URL on completion.
- Get Mystic Task Retrieves the current status and results of a Mystic generation task by task_id. While the task runs the status reads CREATED or IN_PROGRESS; once COMPLETED, the generated array carries the finished image URLs the rest of the flow consumes.
Frequently Asked Questions
What can FlowRunner do with Freepik?
FlowRunner agents can run Search Resources, Get Resource, and Search Icons in Freepik, plus 3 more actions.
Does connecting Freepik to FlowRunner require OAuth?
No. Freepik connects to FlowRunner with an API key, no OAuth flow required.
Can Freepik trigger a FlowRunner workflow automatically?
Freepik doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with Freepik
$100 in credits. No card required. Connect in minutes.