Are AI-Generated Presentations Good Enough for Real Client Work?
*This is independent WhatAI editorial coverage of a RoboNuggets video. Jay E and RoboNuggets have not endorsed or sponsored this post.*
Jay E from RoboNuggets demonstrates a workflow using [Claude Code](/tool/claude) and MARP to generate presentations through Markdown and code-based templates.
**Why code-based presentations are different**
MARP allows a presentation to be written in Markdown and rendered through a repeatable visual system. Claude Code can help draft content, organise the slide sequence, and update files. Instead of designing each slide by hand, the author edits text and lets the template control the visuals.
**The main advantage may be consistency**
A code-based template can preserve fonts, spacing, structure, and recurring elements across many decks. For a team that produces the same report every week or month, this may be more useful than producing one fast draft. Updates become text edits, and version control works the same way it does for code.
**What the workflow may not solve**
- Weak reasoning
- Generic writing
- Poor evidence
- Slides with too much text
- Custom art direction
A presentation succeeds when it helps an audience follow an argument. AI can accelerate the draft, but speed does not guarantee clarity. A deck for a high-stakes client meeting usually still needs an editorial pass and, in many cases, a design pass.
**The question for the WhatAI community**
**Would you send an AI-generated MARP presentation directly to a client, or would it still require a full design and editorial pass?**
For anyone using AI in their deck workflow:
- Which part saves real time: the writing, the structure, or the formatting?
- Have clients noticed, positively or negatively?
- Does a Markdown-based system beat a slide tool with AI features built in?