Flo

How do you make AI-designed emails look on brand?

By the Flo team · Updated

Short answer

Give the AI the brand's real assets and rules, not a description of them. That means the actual logo, the brand's web fonts, its exact colours, real product and lifestyle photos, its voice, and its standing rules. Then every email starts from that context, and each piece of feedback becomes a rule so the next email gets it right.

Why AI emails look generic

When AI email output looks off brand, it is rarely the model's taste. It is missing inputs. A model told "the brand uses a warm, minimal look" invents its own version of warm and minimal. A model given the real logo, fonts, colours and photos uses them.

The inputs that matter most

  1. The real logo, in versions for light and dark backgrounds.
  2. The brand's fonts, read from its website, with close fallbacks.
  3. Exact colours, including button and background colours from the site.
  4. Real product photos, current packaging, most popular products first.
  5. Lifestyle imagery from the brand's own site and social.
  6. Voice: how the brand writes, words it uses and avoids.
  7. Standing rules: every "always" and "never" the client has said.

Rules are what make it stick

The difference between a tool that is on brand once and one that stays on brand is memory. When the client says "never put text over the product" or "the discount code is always bold", that needs to be saved as a rule for that brand and applied to every future email, without anyone repeating it.

Keep a series consistent

For a launch or sale series, lock the shared parts across every email: headline typeface, discount code style, offer line, button style and section headings. Vary the hero and layout so each email feels fresh. AI is good at this when the lock is explicit. See can AI design a whole Black Friday series?

Test with a hard brand

Any tool looks good on a brand with bold colours and great photography. Test on a client with a subtle look, a tricky logo or a strict guideline. That is where the brand inputs show.

How Flo stays on brand

Flo reads each brand from its website: the logo, fonts, colours, real product photos and wording. See how AI designs an email from a website. Each brand keeps its own memory of rules, and feedback can be saved as a rule so it applies next time. Series share a locked look. Flo never swaps a real product photo for a made-up one.

Questions agencies also ask

Can AI use a brand's custom fonts in emails?
Yes, if it reads the brand's web fonts. Text set in images shows the exact font in every inbox; live text needs a close fallback for inboxes that block web fonts.
Does AI need brand guidelines to design on brand emails?
Guidelines help, but most DTC brands do not have complete ones. The brand's website and best past emails cover most of what is needed.
How do I stop AI repeating the same mistake for a client?
Save the feedback as a rule for that brand, so every future email starts from it.