Practical AI Workflows (Part 2)

The last post covered two workflows (one for students, one for professionals); here are three more workflows for creators grinding through content, researchers buried in papers, and founders trying to do six jobs at once. Same idea though: small, repeatable systems that quietly take something off your plate.

---

1. For Creators: Kill the Content Blank Page With Claude

The real enemy of consistent content creation, is the gap between having an idea and actually having something to post.

You think of something interesting on a walk, you tell yourself you'll write it up later. Later arrives, and the thought is gone, or worse, it's still there but flat. You open a blank document, and the idea feels obvious, boring, not worth anyone's time. So you write nothing.

This is a production problem, not a creativity problem.

The Tool: Claude (claude.ai)

Keep a running note (your phone's default notes app is fine) where you dump raw thoughts as they happen. One line is enough. "The reason most people's productivity advice doesn't work in Nigeria" or "Why I think AI tools are being marketed wrong."

At the end of the week, open Claude and paste this:

I'm a [creator type e.g., LinkedIn content writer, newsletter writer, YouTube creator].
My audience is [describe them briefly].
My tone is [e.g., practical, conversational, no corporate fluff].

Here are 5 rough ideas I captured this week:
[paste your notes]

For each idea:
1. Tell me which has the most potential and why
2. Give me a 3-point outline for the strongest one
3. Write a strong opening hook for it

You'll get an outline, a hook, and a clear sense of which idea is actually worth developing. The blank page problem disappears because you're not starting from nothing. What used to take 45 minutes of false starts and staring takes about 10, you now spend the rest of your time writing with structure instead of wandering.

---

2. For Researchers: Process Papers Faster With Perplexity + Claude

Most of the time you spend "reading papers" is actually spent trying to figure out if the paper is even worth reading. You open the PDF, scroll through the abstract, skim for methods, get lost in the literature review, and surface 20 minutes later not entirely sure what you just read.

Multiply that by 15 papers and you've lost a working day to orientation.

The Tools: Perplexity AI + Claude

Step 1 Filter fast with Perplexity. Before downloading anything, ask Perplexity: "What are the 5 most cited recent papers on [your topic], and what is each paper's core argument?" You'll get a sourced overview that helps you identify which papers actually belong in your reading list and which ones are peripheral.

Step 2 Extract, don't just read. When you open a paper, don't read it linearly. Copy the abstract, introduction, and conclusion into Claude with this prompt:

I'm researching [your topic]. Here is the abstract, introduction, and conclusion of a paper I'm reviewing.

[Paste text]

Please give me:
1. The core argument in one sentence
2. The methodology in plain language
3. The key finding
4. One limitation or gap the paper itself acknowledges
5. How this might connect to [related concept or your research question]

What you get back is a structured brief like a research assistant read the paper and reported back to you. You can then decide in 90 seconds whether to read the full paper or move on.

Step 3 Build a synthesis. After processing 8–10 papers this way, paste your Claude summaries back into a new conversation: "Here are summaries of 10 papers I've reviewed on [topic]. What are the common themes, tensions, and gaps across them?"

You now have the backbone of a literature review, or at minimum a clear map of the research landscape. This reduces the time to assess a paper from 20+ minutes to under 5. Across a research sprint, that's hours recovered.

---

3. For Startup Founders: Build Your Weekly Thinking Partner With Claude

Founders are constantly context-switching. One hour you're thinking about product, the next you're in a sales call, the next you're trying to review a contract, the next someone wants feedback on a design. By Thursday you've touched fifteen different problems and made real progress on maybe three.

The issue isn't effort. It's that there's no system to process decisions quickly and move on.

The Tool: Claude (claude.ai)

How to solve this problems:

Once a week (Sunday evening or Monday morning, pick one) open a Claude conversation and paste this framework:

I'm the founder of [one-line description of your startup].
We're currently at [stage, e.g., pre-revenue, post-launch, Series A].
My biggest focus this week is [one thing].

Here are the 3 decisions or problems I need to think through this week:

Problem 1: [Describe it in 2–3 sentences]
Problem 2: [Describe it in 2–3 sentences]
Problem 3: [Describe it in 2–3 sentences]

For each problem:
- Ask me the 2 questions I most need to answer before making a decision
- Give me one framework or mental model that applies
- Flag any assumption I might be making that's worth pressure-testing

What comes back is structure provided by Claude. Suddenly a murky problem has two clear questions underneath it, and a framework you can apply, that's what you actually needed. This cuts the "what do I actually do with this problem" spin cycle from hours to minutes. More importantly, it prevents you from carrying half-processed decisions in your head all week.

Bonus tip: When a decision is made, paste a quick summary of the outcome back into Claude and ask it to write up a one-paragraph decision log: what you decided, why, and what you'd need to see to revisit it. Do this consistently and you build an invaluable record of how your thinking evolved.

---

The pattern across all three of these is the same. You're not asking AI to think for you. You're using it to clear the friction between having a problem and making progress on it.

That's the only way these tools actually earn their place in your day.

So, which one fits the problem you're dealing with right now?