TLDR: AI content often sounds generic because inputs and prompts are too generic, and not specific to your brand. FireBelly founder Mary-Anne Danarzo fixes this with an “AI brand brain”, a closed AI environment fed only your own brand documentation, from customer avatars to a 35 page tone of voice guide. The same rule drives her CRO work – specific beats generic, every time.
On this episode of Leaders on Shopify, we sat down with Mary-Anne Danarzo, founder of ecommerce agency FireBelly.
She spoke about a business problem she sees repeatedly, not from AI itself but from scaling brands lacking cohesive direction. A solution to this is an AI brand brain: a method for getting AI to produce genuinely on-brand content instead of the generic output most teams settle for.
We’re sharing her framework here as it lines up with what we believe about AI and data. Data only helps if it’s specific and clean enough to act on. Mary-Anne applies that same principle to brand and creative, not just reporting.
What is an AI brand brain?
Most teams give AI a prompt and a couple of adjectives. “Friendly, collaborative, no jargon.” And then wonder why the output reads like every other brand’s AI content.
Mary-Anne’s diagnosis is that without hard constraints, it defaults to generic styling. Her fix isn’t smarter prompting, it’s a closed, internal environment where AI can only reference documentation you’ve given to it, what she calls a brand brain.
What should go into an AI brand brain?
As per Mary-Anne, FireBelly’s brand brains include:
- Company vision, mission, and values
- Detailed customer avatars: who buys, and why – an ideal customer profile
- A tone-of-voice document she says runs to roughly 35 pages, with specific phrases and emojis to use and avoid
- Tone of voice broken down to individual SKU level
She’s upfront that building one is tedious and takes time. Her words: “It’s boring as hell.” But it’s the input that makes AI-generated ad copy and email campaigns usable without a rewrite.
The same logic behind the conversion rate stat
This is where Mary-Anne’s brand work and her CRO work connect. She told us that most brands treat conversion rate optimization as a one-off project, not an ongoing discipline, and most stop at surface fixes like checkout flow.
Her actual method is to look at conversion page by page, not as a site average, and find the pages driving 80% of revenue. She pairs that with direct customer research. At one point, while working in retail, she spent ninety minutes standing in a Tesco aisle watching how shoppers browsed power banks before a product relaunch.
Same principle both times – specificity beats generic effort. A two line tone of voice slide underperforms a 35 page document, the same way a site wide conversion average underperforms specific page level analysis. It’s a version of the same discipline we push on the reporting side, just applied to brand and creative instead of ad spend.
Building AI product content at scale
Mary-Anne extended the idea to visual content too. For WRE, an athleisure brand FireBelly works with, the shift has been from shooting every color and setting separately to shooting one product library, then letting AI generate variations from that base.
Her recommended tool for this on Shopify is Tolstoy, which connects to a store’s product catalog and lets brands build consistent AI avatars, fixed AI generated models used across a site instead of one-off images that drift in style.
AI tool recommendations
Straight from the conversation with Mary-Anne, these are her picks:
- Lifetimely for lifetime value
- Skio for subscriptions
- Okendo for reviews
- Tolstoy for AI generated product creative
Listen to the full conversation
Mary-Anne Danarzo is the founder of FireBelly. Find her on social under the same handle: Mary-Anne Danarzo. The full episode of Leaders on Shopify is linked here.
