Amanda AI

Amanda is arrobaMail's AI agent. She works like a strategist: she analyzes your audience and your campaigns, plans, generates the content, and leaves you prioritized recommendations. You work with her through chat, both on the site and inside the platform.

Analyze a campaign across axes (deliverability, engagement, content quality, and list quality) and give it a 0–100 score with prioritized recommendations.

Generate full campaigns from a single instruction: copy, HTML, subject line, and preheader.

Learn your brand identity from your website or previous campaigns (Brand Kit) and apply it to every generation.

Create images for your sends, suggest tone improvements, and translate — all inside the editor.

It's Amanda's unit of use: every complete operation — creating a campaign, requesting a change in chat, generating a subject line and preheader, setting up the Brand Kit, or analyzing a send — consumes 1 generation. Plans include a monthly quota that renews each cycle; unused generations don't roll over. Tip: the more complete your initial instruction, the fewer generations you'll spend on corrections.

Yes. Amanda builds a Brand Kit with your identity (she can generate it on her own from your website URL or from a previous campaign) and applies it to every generation, so the content comes out in your voice instead of a generic tone. And you can freely edit any result.

No. Amanda plans and drafts, but no campaign is sent until you review and approve it. Control is always yours — and the same applies when you connect an external assistant via MCP.

No. Amanda contributes analysis, planning, and production so your team can work faster and with better information, but strategic decisions and final judgment stay human. She's a strategist who assists, not a replacement.

She understands and responds in Argentine Spanish, neutral Spanish, English, and Portuguese. She adapts to your account's language and the brand tone you've configured.

Starting 7 days after the send: opens and clicks keep coming in during that period, and the analysis needs representative data to be useful. Running it consumes 1 generation; follow-up questions in the analysis chat do too.