Welcome to the 9th edition of Work in Beta.

In this edition, we help you reimagine your workflow when creating Presentations (Decks) with AI. Trying to do it like it has been done for decades might be futile and our Two-Step Model helps with exactly how to augment the workflow.

We are also hosting our third AI WORKSHOP for non-technical professionals this month on 14th March 2026 and would love to have you join. The details are below in this email.

So, let’s dive in!

When you make a presentation manually, content and design happen together. You type a headline, adjust the font, add a bullet, move a text box - all on the same slide, at the same time. That works fine when a human is doing both jobs. There's no rule that says you must do one before the other.

But when you bring AI into the process, that changes.

Ask any AI tool to "make me a 10-slide deck about Q4 results" and you'll get a partial job on the content and a partial job on the design. The narrative will be generic. The slides will be functional but forgettable. That's because generating content and generating design in one prompt is a lot to handle in one go - the AI compromises on both.

The fix: stop asking AI to do both at once. Break it into two steps.

As of March 2026, every major AI tool can generate presentations. But to move away from creating generic presentations to ones that actually do the job right for you, you need to work with what we call it the Two-Step Model.

Step 1 - Content & Narrative Architecture. This is where you do the thinking with AI. You discuss, riff, structure, and go back and forth on what exactly to say, in what order, for maximum impact. You're not touching design yet. You are building the argument.

Step 2 - Design & Visual Language. Once content is locked, you move to design. This step uses a theme, template, or brand guideline you already have. The AI applies your visual language to the structure from Step 1. When Step 2 is done, you're 60-70% of the way to a finished deck.

That's the workflow. Step 1 flows into Step 2. You go from "nothing" to "mostly done" and the remaining 20-30% is your editorial pass: refining the narrative, adjusting emphasis, adding your voice.

Why two steps instead of one? Separating the two steps means each step gets full attention and you get a first draft that's actually closer to what you need.

The skill here: thinking before prompting.

Most people prompt with: "Make me a 10-slide deck about market trends." That's a lazy brief. You'll get a lazy deck. The workflow starts before you touch any tool.

What good content thinking looks like:

Define the purpose clearly:

Choose a narrative structure:

Define the skeleton:

We've distilled this into a reusable framework. Before prompting any AI for a presentation, answer these 6 questions:

Feed these answers to any LLM. The better your brief, the better your first draft. Every time.

The prompt template - copy this, fill in the brackets, paste it into any AI tool:

The difference this makes: "Make a 10-slide deck on market trends" produces generic garbage. The same tool with the structured brief above produces a focused, narrative-driven first draft. Same AI. Wildly different output. The workflow is the variable, not the tool.

For high-stakes presentations — board decks, client pitches, investor updates - Step 1 is exactly where the Explorer → Writer → Auditor method applies.

That three-chat method is Step 1 for presentations that need to survive scrutiny.

The skill here: having design intent without being a designer.

You don't need design skills. But you need to tell the tool what "good" looks like for your context.

Do you have a template?

What visual density fits the audience?

The prompt template for Step 2 - once your content is locked from Step 1, paste this into the same or a new chat:

Key insight: The best first drafts come from tools that know your template. If brand compliance matters, start with your template loaded. If speed matters more, use the connector/design tool path.

Once Step 2 is done, you're looking at a 60-70% finished deck. The structure and narrative flow will be solid if your Step 1 input was good. The design will match your template if you provided one, or be functional-but-basic if you didn't. Your job from here is editorial - refine the narrative, adjust emphasis, add your voice. The time you saved is the formatting, structuring, and blank-slide-staring time. The time you invest is the thinking time. That's the right trade.

The Two-Step Model works across every major AI ecosystem. Here's where to start:

The tool doesn't matter. The two-step workflow does. Pick whichever ecosystem you already use and run it.

The first draft is no longer the hard part. The thinking is.

The Two-Step Model doesn't make presentations easier to care about. It makes the mechanical parts disappear so you can focus on what actually matters: what you're saying, to whom, and why.

The skill is Step 1. The tool is Step 2. The workflow connects them.

Pick one presentation. Run the workflow. See what happens.

Over the last 2 editions, we have already helped participants from diverse functions including engineering, consulting, marketing, operations, product management, investment banking and more, to get more efficient at work using AI.

We specifically focus on:

Feedback we have received:

If you wish to join the next session on 14th MARCH 2026, fill this form and we will reach out to you with details: https://forms.gle/D6TGEVoFBUqKPsP49