Ahead with AI #2: Research on autopilot, AI video briefs, and smarter prompts

Apply AI in 5 minutes every Sunday: 3 prompts, 2 ways to use AI, 1 key insight

Stay ahead of the curve by turning AI from something you read about into something you use.

In today’s Ahead with AI:

• Prompt: get one playful challenge outside your comfort zone

• Prompt: turn your top-performing content into 10 new posts

• Prompt: reframe problems when you feel stuck

• How to: schedule a recurrent research task in Perplexity

• How to: create video presentations with NotebookLM

• Key insight: an easier way to give AI the context it needs for better outputs

---

3 prompts to kick off your week

There are endless ways to use AI, but many of them slip by because we never think to ask. Here are three to try this week with your favorite model. Add your own context so the answers are useful.

These are ready to go, but if you want prompts tailored to your own needs, try my Prompt Generator.

I. For you: One Fun Leap past Comfort

Based on everything you know about me from your memory and our past conversations, give me one small but unconventional challenge I can try this week that pushes me outside my comfort zone, while still feeling playful and enjoyable rather than forced.

Explain:

• Why you chose this specific challenge for me (based on my patterns, tendencies, and interests).

• What creative or emotional benefit it could unlock if I engage with it fully.

• How I should approach it so it stays fun and doesn’t turn into pressure.

Constraints: Avoid clichés. Suggest something unexpected, practical, and tailored to me.

Here’s what it suggested for me. Fun enough that I’m trying it.

---

II. For work: Replicate Your Top-Performing Content

Analyze my [type of content] that’s been getting the most traction.

Step 1: Break down what’s working: structure, phrasing, tone, emotional triggers, timing, platform mechanics.

Step 2: Now, create [10] new [type of posts] for me. Each one should stay true to my [brand/personal] voice while following the viral patterns you identified. Make sure every piece can stand fully on its own without needing extra explanation, and that it’s tailored to perform well on [platform].

Step 3: For each piece you generate, include the content itself, followed by a single sentence explaining why it works. Keep this explanation focused on elements like clarity, emotional pull, or how well it fits the platform. Also, tag each piece with its primary trigger (for example: curiosity, utility, status, or novelty).

Step 4: Once all 10 pieces are complete, rank them from strongest to weakest so I can see which have the highest viral potential. Then, step back and highlight the top three common patterns that showed up across these best-performing pieces. This will give me a clear sense of what consistently works and can be reused going forward.

Constraints: Be specific and evidence-based. No generic advice. If a choice comes from my past winners, say which pattern it mirrors.

Context:

• Audience details: [add context]

• Platform: [add platform]

• Content type: [e.g., hooks, carousels, notes, posts, tweets]

• Past posts: [paste posts or upload a doc]

---

III. For perspective: Challenge Your Framing

Help me think through this situation: [add context]

First, analyze how I’m currently framing this. Point out assumptions I might be making, missing context, or where emotion could be leading the way.

Then, ask me 3–5 clarifying questions that push me to see this from different angles. Challenge me where I seem too certain or too narrow.

Finally, suggest 2–3 ways I could reframe the problem so I can move forward with clearer thinking.

Constraints: Be logical, specific, and critical. Don’t just agree with me, help me see what I’d normally overlook.

---

2 ways to put AI to use

AI isn’t measured by what it promises, but by what it does. Here are two ways you can put it to work this week.

I. Schedule a recurrent research task in Perplexity

This guide shows you how to use Perplexity to set up automated research tasks.

How it works:

1. Open Perplexity and log in or sign up.

2. In the search bar, enter a query like:

   Summarize all newly posted research papers on large language models (LLMs) from the past 7 days across arXiv, OpenReview, Harvard, MIT, Google Scholar, SSRN, Frontiers, ResearchGate, and recent major venues (NeurIPS, ICLR, ICML, ACL, EMNLP). For each paper, include the title, a 2–3 sentence plain-language summary of the main contribution or discovery, why it matters, the practical implications and takeaways, and a direct link to the full text. Group related papers by theme (e.g., reasoning, alignment, efficiency, multimodality, evaluation, hallucinations, creativity, AI in the workplace etc.), note any new datasets or benchmarks introduced, and finish with a brief synthesis highlighting the most impactful ideas and 2–3 open questions to watch next.

3. Click the three horizontal dots in the top right corner and select Create task.

4. Set the frequency, day, and time when the task should run, and how you want to be notified when it’s done.

5. Under “Sources”, select Web + Academic for this type of query. (If you were tracking Reddit threads, you’d choose Social only.)

6. Optionally, choose Deep Research instead of Fast Search for more detailed results.

Where it’s useful:

• Stay compliant: track new regulations, policies, or industry standards as they emerge

• Stay competitive: monitor competitor moves, product launches, and pricing changes

• Stay informed: follow new research, publications, or expert debates without manual searching

• Stay connected: capture customer discussions and feedback from forums and social platforms

• Stay curious: build weekly digests on any personal topic of interest so you never miss what matters to you

Estimated time:

5 minutes

Cost:

$20/month (Pro plan required for Task automation)

💡 Perk: You can currently get a free one-year subscription if you sign up with your PayPal account

---

II. Create a video presentation with NotebookLM

This guide shows you how to turn your own resources into an AI-narrated video presentation with NotebookLM.

How it works:

1. Log in or sign up to NotebookLM with your Google account.

2. Click Create new and add your sources: PDFs, Google Docs, Google Slides, web links, YouTube links, or pasted text.

3. Add a title to your notebook so it’s easy to find later.

4. In the right panel, click Video overview and wait for it to generate.

5. Review the result (here’s the one I generated), then share the link from NotebookLM or download the video.

Where it’s useful:

• Turn research papers, reports, articles, courses, or even books into narrated video briefings you can share with your team and others

• Repurpose slide decks, notes, or transcripts into engaging video explainers for training or workshops

• Transform YouTube links, PDFs, or web content into quick, skimmable video overviews for learning or review

Estimated time:

5 minutes (just a few clicks to set up and add your resources)

Cost:

Free plan available (with limitations)

---

1 key insight

One idea, lesson, or finding I found useful (or thought-provoking) enough to share.

I. The simplest way to improve your AI outputs: add context

The biggest factor in getting useful answers from an LLM is the context you give it. Without context, it guesses and produces generic outputs. With context, it reasons and gives answers tailored to your needs and situations.

One of the easiest ways I’ve found to make sure you add enough context every time is to stop typing and start talking. The voice feature lets you explain your ideas as if you were speaking to a colleague. Quicker, more natural, and with the details you’d otherwise leave out.

Most major LLMs now support voice, and it’s one of the simplest upgrades you can make to your workflow.

---

---

Your turn

Tried any of these (or something else with AI) this week?

Hit reply or drop a comment to share how you used it. I love hearing how you’re putting this into practice.

PS: I might feature your example in a future issue.

---

From the notebook

In case you missed these, here are a few of my recent deep dives:

• Good vs. bad prompts: side-by-side outputs showing how strong prompts create 30x better results + a tool to make your own

• Free Gmail automation: keep your inbox clean and get one daily digest

• Catching my biases with ChatGPT: how I use it to spot blind spots

• Ahead with AI #1: voice cloning, kids’ storybooks, and more

---

Work with me

• Consulting: I’ll help you integrate AI into your workflow and create systems that save time, reduce friction, and scale your results. If you’d like to explore working together, just hit reply.

• Paid partnership: I’ll promote your product to +2,600 AI-curious readers, through either direct ad placements or hands-on tutorials and workflows.

---