Hey there,

Last week I had a proposal to build for a prospect.

Normally, this is where I'd open PowerPoint, stare at a blank template for too long, start moving boxes around, and slowly piece together something that looks decent but took way longer than it should have.

This time, I opened Claude.

I uploaded my deck template and fed it all the context: who the prospect is, what my plan was for the proposal, what I wanted to highlight. I asked it to think through the presentation with me.

Before Claude even built any content, it proposed a structured outline of what the deck should cover, how it should flow, and what to emphasize. I read through it and thought... I actually agree with all of this.

But I was curious. I'd also been hearing about people building decks with Manus. So, I took that same outline and uploaded it to Manus to see what a different AI tool would produce.

I got two similar outputs, threw them together in a single deck, and went through and chose which versions I liked better, or which concepts I liked from each. About 80% of the time, I preferred Claude's output. About 20%, Manus.. but having both versions gave me more raw material to work with.

It still wasn't 100% there, so I typed out every note I had — design tweaks, content changes, things I liked, things I didn't — and fed it all into Claude's PowerPoint plugin.

Then I watched it make the changes.

Was the final deck perfect straight out of AI? No. Here's are a few areas I spent time directly in PowerPoint:

Text edits. Small wording tweaks that needed a human eye. Easier hcanges to make directly rather than through AI.

Logos and headshots. AI was terrible at this. At one point, it tried to recreate a client's logo by typing the company name in a different font. Not quite the same thing...

Knowing when to stop asking AI. There were moments where I spent more time describing what I wanted AI to change than it would've taken to just do it myself.

The "AI design" problem. AI makes slides differently than a human would. So many unnecessary boxes and containers. Layouts that look good but are annoying to edit after the fact.

Net-net? It was faster. The output was arguably better than what I would've built solo.

The AI helped me work through what to say and how to say it before I ever thought about slide layouts.

I'm excited about what this means for how we build proposals going forward. Less time wrestling with slides. More time thinking about how we actually want to communicate.

Enjoy!

Shaalin

P.S. Have you tried anything like this recently? If so, reply and let me know!