The brand checker is an AI agent that helps you answer a simple question: “Is this on brand?” Upload an asset in your brand’s Guidelines portal and get a list of what is off brand, why, how confident the agent is, and how to fix it.
Good to know: Availability, usage limits, and supported file types vary during the beta rollout.
What is the brand checker?
The brand checker puts a first brand check where your rules already live: in Frontify.
You upload an asset such as an image, and the agent compares it with the checks your brand team has defined and the reference material already maintained in Frontify.
The goal is simple: help people catch brand issues while the work is still in their hands, before it reaches a reviewer or gets shared more widely.
What does it check?
The brand checker does not use a fixed list of checks.
Instead, your brand editors define checks in plain language, such as:
“Check fonts”
“Review the colors”
“Verify logo usage”
“Check spelling and tone”
The agent then works out which brand rules apply.
What the brand checker uses | What it means |
Brand rules | Written rules maintained in your Guidelines |
Brand Essentials | Approved fonts and color palette |
Logo library | Approved logos maintained in Frontify |
Brand conventions | General conventions that can be answered from the asset and your brand references alone |
This means your brand team defines what “on brand” means, while practitioners simply upload the work they want to check.
How to enable the brand checker
The brand checker needs to be set up once by a brand Owner or Editor before people can use it.
Before you start
Make sure your brand has:
Brand assistant enabled — this is currently required for the brand checker to work.
Essentials with your approved fonts and colors.
A logo library in the same brand if you want to check logo usage.
A submission library where assets submitted for checking can be stored. This is currently required during beta.
Set up the brand checker
Open your brand’s Guidelines portal and add the brand checker custom block.
Open the brand checker and select Configure.
If prompted, enable brand assistant, then return to the brand checker. You may need to close and reopen the page.
Select the submission library where uploaded assets should be stored.
Configure the checks you want the brand checker to perform. You can use checks for logos, colors, typography, layout, and writing, or create your own custom checks.
Edit the check instructions to make them specific to your brand. For example, you could extend a writing check to include grammar and spelling.
Save your configuration.
Once configured, the brand checker is ready to use. Practitioners can upload an asset and run the checks without configuring them individually.
Beta note: The setup experience is still evolving. Assets submitted for checking are currently stored in the selected submission library, and some setup requirements may change during the beta.
How to use the brand checker
Once your brand team has configured the checks, using the brand checker is straightforward:
Open your brand’s Guidelines portal.
Upload the asset you want to check.
Run the brand checker.
Review the results and make any relevant changes.
Continue with your normal review or publishing workflow.
You do not need to select individual checks for each asset. The brand team defines the checks once, and those checks are applied to the assets practitioners bring to the tool.
What happens when you run a check?
The brand checker returns a result for each check it can evaluate.
Results include:
Result detail | What it means |
Pass or Finding | A potential issue where the asset does not appear to follow the relevant brand standard |
Reason | The rule or evidence that led to the result |
Confidence level | High or Low confidence in the result |
Suggested fix (if finding) | A practical change to make when the agent can provide one |
For example, instead of simply telling you that an asset has a typography issue, the result can explain what it found, why it matters, and what you can change.
Important: A Low-confidence finding is worth a second look rather than something to treat as a definitive answer.
What you do vs. what the brand checker does
What you do | What the brand checker does |
Upload an asset | Reads the asset and runs the checks defined by your brand team |
Review the results | Shows passes and findings, with the reason, confidence level, and a suggested fix where there is one |
Make changes | Helps you correct the asset before it reaches a reviewer or is shared |
What does the brand checker use as its source of truth?
The brand checker reads your existing Frontify content.
Guidelines provide the written brand rules.
Essentials provides approved fonts and color palettes.
Your logo library provides approved logos.
When those references are updated, the next check reflects the updated information.
This means there is no separate rulebook that your team needs to export, upload, and maintain.
What is currently supported?
The brand checker is designed to answer whether an asset follows a brand standard that can be evaluated from the asset and your brand’s own references.
It can help with checks such as:
Typography
Colors
Logo usage
Spelling and tone
Other written brand standards defined by your brand team
The quality of the results depends on how clearly and thoroughly the relevant brand rules and reference information are defined in Frontify.
Current limitations
The brand checker is evolving through beta, so some capabilities are intentionally limited.
Currently:
Video, motion graphics, and audio are not supported.
Non-image docs: PDFs and PPTs
Checks that require external sources, live web pages, legal or trademark databases, or other external tools are out of scope.
The brand checker does not certify accessibility compliance. It can measure a contrast ratio against a required minimum, but it does not produce WCAG certification, a VPAT, or an ACR report.
Checks run on demand after you upload an asset. The tool does not provide live feedback while you edit.
A separate “specs file” for additional review procedures is not currently supported.
Important: The brand checker is designed to answer whether an asset follows your brand standard. It does not judge whether the work is effective, tasteful, original, or on trend.
Getting the best results
The brand checker is only as good as the brand rules and reference material it can read.
For the best results, keep your:
Guidelines
Brand Essentials
Logo library
are complete and up to date. If a rule or reference is missing, the Brand checker may not have enough information to perform the relevant check. Rules that are shown only in images (e.g., logo spacing or color rules) may also not be reliably evaluated.
Frequently asked questions
Does the brand checker respect my Frontify permissions?
Yes. The brand checker respects the permissions and targets that apply to you. It only uses brand information you have permission to access. For example, if you can access the logo library but not a writing guideline, the checker can use the logo information but will not use or expose the restricted writing guideline.What happens if the brand checker is not sure?
The brand checker can return a Low-confidence result when the signal is less certain. If there is not enough information to evaluate a check, the result can be unassessed rather than guessed. Treat Low-confidence or unassessed results as a reason for human review.Does the brand checker replace a brand reviewer?
No. The brand checker is an advisory tool. It helps catch issues against defined brand standards, but it does not approve or sign off work. Human reviewers remain responsible for creative judgment and final decisions.How is the brand checker different from using Frontify MCP with an AI assistant?
Frontify MCP lets an AI assistant access your brand knowledge and use it when answering questions or giving an opinion. The brand checker is designed for repeatable, defined checks against an asset. Your brand team defines the checks once, and the checker evaluates assets against those checks using the relevant brand references in Frontify.What AI technology powers the brand checker?
The brand checker uses a combination of technologies. An LLM handles natural-language understanding and rule orchestration, while Frontify’s in-house computer vision handles visual checks such as exact colors, typography, and logos. AI processing uses an OpenAI GPT model hosted on Microsoft Azure in the EU and accessed through OpenRouter.Is my content used to train AI models?
No. Customer assets, guidelines, and other customer content processed by the brand checker are not used by Frontify or the model providers to train or fine-tune AI models.Where is my data processed?
Frontify runs the brand checker on its AWS infrastructure in your customer region. For EU customers, this is in Frankfurt. AI requests are routed through OpenRouter to Microsoft Azure in the EU. Monitoring traces are stored in Langfuse in Ireland.What data is sent to the AI service?
The brand checker sends the information needed to perform the requested checks. This can include the asset, relevant metadata, check requests, relevant brand information and guideline excerpts, and results from Frontify's visual measurements. Approved logos may also be included where relevant. Font files and the rest of the logo library remain inside Frontify.Can I use the brand checker outside the Guidelines portal?
During the open beta, the consistency checker is planned to be available through API and MCP, allowing it to be called from external tools, internal applications, build pipelines, or AI agents. The same service is intended to use the same brand knowledge and checks.How much does the brand checker cost?
During the closed beta, the brand checker is available within the beta usage limits at no additional cost. Frontify covers the model costs during this phase. At general availability, each account will include a set volume of checks, with additional volume available for purchase.Can I join the beta?
The initial closed beta is invite-only for participating Enterprise accounts. An account administrator must opt in. The beta is intended for testing and learning, so availability, usage limits, and supported capabilities may change as the product evolves.How should I evaluate the brand checker during beta?
A useful approach is to choose a guideline and a small group of assets, then compare the brand checker’s results with decisions made previously by your human reviewers. This helps you understand where the checks are reliable and where human review is still needed.Which languages are supported?
English, German, and French are currently the most reliable languages. Results may be less reliable for less common languages, and typography checks work best with Latin-script typefaces.
A more consistent way to apply your brand
Publishing brand rules is not the same as applying them. The brand checker brings a first-pass check into the work itself, so practitioners can catch issues sooner and brand teams can apply the same standards across more of the work their organization produces.