Every brand drifts.

Not all at once. An accent colour gets eyedropped from a screenshot instead of copied from the kit. A headline is set in three weights because three people touched it. The logo loses a little clear space to fit one more line of copy. Each of these is small enough to miss on its own. Put a campaign's worth of them side by side and the brand starts to look like several brands.

Six assets from one campaign side by side: posters, social posts and banners that no longer quite match — One campaign, six assets, small differences that add up.

Brand review exists to catch this, but it is slow and it lands on the busiest people. A designer or a brand lead checks each asset by eye, against guidelines they mostly carry in their head, usually right before it ships. Two reviewers will flag different things. The same reviewer will flag different things on a Friday afternoon.

Today we are launching Kitana Lite, a brand review agent that does this check for you. She is live in public beta at kitana.replikit.ai.

What Kitana does

Kitana answers one question: does this creative match the brand it is supposed to match?

Give her your brand. Point her at your website, or upload your brand guidelines. She builds a brand kit from it: your colours, your type rules, how your logo is used and the layout patterns you repeat.
Send her your work. Submit up to 10 assets in one review: social posts, banners, posters, landing page screenshots.
Get the report. About two minutes later, a report arrives in your inbox and waits in your review history. It has an overall score from 0 to 100, a score for each part of the brand, and findings for each asset.

A brand kit per brand or sub-brand. Kitana reviews against the one you choose.

What she checks

Every review looks at five parts of your brand:

Colour: your palette, contrast, and any values that are off-palette.
Typography: family, weight, and where each line sits in your type scale.
Logo: placement, clear space, and whether it is used on the right colours.
Layout: spacing and grid, against the template the asset came from.
Coherence: whether a set of assets reads as one campaign. Kitana scores this across the whole set, not asset by asset.

Findings are specific, and each one is pinned to the place on the asset it refers to. On the Summer Frequencies poster at the top of this post, Kitana does not say "the colour feels off". She says the accent is #4F86F7 and the brand accent is #1D4ED8. She says the headline uses three weights, 700, 500 and 600. She says the logo has 0.6× the minimum clear space. Each one is something a designer can fix in a minute, or something an AI agent can fix on its own (more on that in our MCP launch).

The score tells you how much attention a piece needs. Scores fall into five bands: Excellent, Good, Needs Work, At Risk and Critical. The poster scored 68, Needs Work.

The score, the band and the top findings, in one view.

Where she works

Kitana runs the same review wherever your team already works. Every channel uses the same brand kit, the same credits and the same review history.

On the web, at kitana.replikit.ai.
In Figma, through the RepliKit plugin. You review inside the file, before anything is exported.
In Slack, so a review can start and finish in the channel where the work is discussed.
In your AI tools, over MCP. Claude, ChatGPT, Cursor and your terminal can run a review, read the findings and check a fix. Read the MCP launch.

The Kitana panel in the Figma plugin, showing a score of 68 and three findings beside the poster — In Figma, the review sits beside the frame it checks.

Kitana posting a monthly report in a Slack channel: 85, up 4, with colour as the weakest area — In Slack, Kitana posts the report where the team already talks about the work.

What she does not do

We want to be clear about the limits, because a review is only useful if you know what it covers.

Kitana reviews. She does not make or edit creative work.
Her score is design advice, not legal advice. A clean review means the work matches your brand system. It does not check trademark, accessibility or regulatory rules.
She does not train on your work. Your assets and guidelines are used to run your reviews and build your kit, and nothing else. They are never shown to another customer.

Pricing

Free: one brand kit and two reviews, with no card. Each review covers up to 10 assets, and you get the full report. The two reviews are a one-time allowance, and they also work inside the Figma plugin.

RepliKit Pro, $39.99 a month: 20 reviews a month, plus Slack, MCP and the plugin's pre-export brand check. One Pro plan covers both Kitana and the RepliKit Figma plugin, on one credit balance. Extra reviews are $2.50 each.

One review is one credit, whether you send one asset or ten, so send the full set when you can.

Every asset, checked before it ships.

Why we built her

RepliKit started with templates: designers lock the brand rules, and everyone else fills in the content without breaking them. That protects the work made from a template. But a lot of work is not made from a template. It comes from agencies, freelancers, other tools and, more and more, from AI.

Kitana is the check for all of that work, wherever it was made. We think of her as the first layer of your brand's infrastructure: a reviewer who reads your guidelines the same way every time, and is ready whenever the work is.

Run your first review free. Two reviews, no card, and the report is in your inbox in about two minutes.