Enhanced image search automatically interprets what an image depicts and makes it searchable in natural language. It’s powered by AI and helps teams find assets faster while reducing manual tagging work.
Currently in beta.
What is Enhanced image search?
Enhanced image search uses Azure OpenAI to analyze the visual content of images in your Frontify libraries. This allows the system to understand what’s depicted in an image and make it searchable using natural language, even if the asset doesn’t have tags or metadata.
For example: Search “team collaborating at a whiteboard” to instantly find relevant visuals, even if those files weren’t manually described.
Key benefits
Automatic visual understanding without manual metadata entry
Faster discovery through natural language search
Increased asset reuse and brand consistency
Improved search accuracy across libraries
Administrative control to enable or disable the feature
Getting started
Enabling Enhanced image search
Enhanced image search is managed through Account Settings → AI Features → Enhanced image search.
Existing customers:
Receive a 60-day opt-out window before the feature activates automatically
Can test the feature or request early activation during this period
New customers:
The feature is enabled by default
Library creators can disable it for specific libraries, and account admins can manage settings globally
How it works
When enabled, all existing image assets in end user–facing library types are processed automatically.
New uploads are processed on arrival.
Only PNG, JPEG, and WEBP formats are supported. SVGs, videos, and other file types are excluded for now.
The AI generates machine-level visual understanding of each image, which is used to improve search relevance, this information is not visible in the asset metadata or description fields.
Searching with Enhanced image search
Use the existing guideline search bar as usual. Enhanced image search adds an extra layer of intelligence by matching your search terms with the visual content of images, even if those images don’t have manual tags or descriptions.
For best results, try descriptive phrases like:
“people working remotely”
“summer outdoor team event”
“close-up of branded packaging”
Known limitations
Enhanced image search is currently in beta, and there are a few important considerations:
For early testing, start with small libraries (up to 1,000 assets).
Only PNG, JPEG, and WEBP files are supported; SVGs, videos, and other file types aren’t processed.
Supported library types include media, logo, and document libraries (the icon library is excluded since it only allows SVG uploads).
Assets with transparent backgrounds may be interpreted as white or black, which can cause incorrect color detection or blank results for black/white objects.
Use cases
Marketing teams finding campaign imagery faster
Designers browsing concept visuals by theme or emotion
Brand managers ensuring all assets are discoverable, not buried in folders