The AI Daily Brief: “15 Ways I Use AI (And the Models I Use for Each)”

Host: Nathaniel Whittemore (NLW)
Date: July 13, 2025

Overview

In this Long Reads weekend episode, Nathaniel Whittemore (NLW) dives deep into the multitude of ways he personally integrates AI into his daily and professional life, breaking down both specific tools and broader use cases. He not only details his own automation and content workflows but also touches on areas where current AI tools excel, fall short, or continue evolving. The episode is practical, reflective, and offers listeners a granular look at modern AI-assisted work.

Key Discussion Points & Insights

1. Superintelligent’s Agent Readiness Audits

Timestamps: [03:23]

• NLW begins by highlighting his company’s custom voice agent (built on OpenAI API) that interviews companies to identify valuable agent use cases for their business.
  • “Voice agents are of course at the very beginning of their mainstreaming…” [04:24]

2. Podcast Production & Editing

Timestamps: [05:01–11:30]

• Descript: NLW's team does all podcast editing in Descript, leveraging its collaborative platform and its AI suite, Underlord.
  • Features used: Automated removal of filler words, removal of retakes, “Eye Contact” deepfake function to fix eye gaze.
  • Memorable moment: “Honestly, it's a testament to how good some of these editing features are that I put up with [Descript] dropping an entire episode every couple of months…” [07:04]
• VEED: Used for rapid review/edit at 2x or 3x speed.
  • “With VEED, you can...review not just at 1x but at 2x or even 3x speed...which is an absolute godsend.” [08:19]
• Short-Form Content (Should-Use): While not yet in active use, NLW recognizes the value of auto-clipping tools like Opus for social-ready short videos.
  • “I should be using...automatic clipping tools...for TikTok, Instagram reels, and YouTube shorts.” [09:38]

3. AI-Aided Writing: Podcast Descriptions and Show Notes

Timestamps: [11:31–14:27]

• Process: Uses AI to turn transcripts into show notes; trusts AI with descriptions but not titles.
  • Often prefers a “single clear determined statement” over formulaic, colon-heavy AI-generated titles.
  • “ChatGPT has a way over-adherence to conventions...which just do not perform well in practice.” [13:08]
• Models Used: ChatGPT 4o or 4.5 Research Preview for this workflow.

4. Long-Form Writing & Newsletters

Timestamps: [14:28–20:48]

• Claude (Opus 4 and latest Anthropic models): NLW's go-to for most longer-form writing or newsletter work, especially when providing style templates and iterative guidance.
  • “You can create custom writing styles that you can come back to over and over again. These are akin to something like a midjourney style template, but for writing.” [17:45]
• Examples of Custom Styles:
  • Tech Translator: “Deliver analytical insights through conversational and authoritative communication.”
  • Cheeky Crime (for his wife’s true crime podcast): “Explore dark topics through sophisticated, witty and playful narrative commentary.”
  • Storyteller's Lens: Journalistic-tone variant for nuanced subjects.
• Use case distinction: Personal LinkedIn/blog posts are still written by hand.

5. Advanced Research Workflows

Timestamps: [20:49–28:15]

• Deep Research (O3/ChatGPT): Used for comprehensive backgrounders, pitch arguments, and strategy preparation.
• Mini Research (Perplexity): For quick fact-checks, daily questions, or lightweight research.
  • “If I'm ever looking for random PE ratio research...I'm going to go to Perplexity for that.” [26:02]

6. Strategic Planning & Ideation

Timestamps: [28:16–33:49]

• Treats O3 (OpenAI’s v3) as an “actual strategic collaborator,” brainstorming and stress-testing both business and speculative ideas.
  • “At this point I talk about pretty much every different strategic idea that I've had with O3 in some way...” [29:11]
• Grok4: Currently parallel testing as a strategic assistant.
  • “I am...giving Grok4 a try as a competitor in this role for at least a week or two.” [32:41]
• Deck/Memo Improvements: Uses LLM reasoning models to tighten pitches, especially for memo outlines, but notes that decks can feel formulaic.
  • Pro-tip: Always push back against “sycophancy” in model outputs to extract better, less formulaic results.

7. Voice Recognition Alternatives

Timestamps: [33:50–35:30]

• Flow/WhisperFlow (WISPR): Game changer for voice input on iPhone, vastly superior to Apple’s built-in solution.
  • “If you use an iPhone...you'll note just how unfathomably, offensively bad its voice recognition software is at this point. ChatGPT is really good, but...many people use Flow or WhisperFlow...total game changer.” [34:18]

8. Visual Creation: Cover Art & Image Generation

Timestamps: [35:31–40:15]

• Ideogram: For episode covers and collateral, especially when fidelity to prompt (especially text) matters.
  • “Ideogram...one: high understanding and high fidelity to what you're trying to accomplish...two: it’s great with text.” [36:08]
• Midjourney: For art experiments, creative backgrounds, slides, or “sheer fun.”
  • “I still love midjourney. It consistently produces just the most interesting and visually arresting things.” [38:29]
• Prompting: Ideogram’s extended auto-generated prompts help remix results creatively.

9. Vibe Coding

Timestamps: [40:16–42:33]

• Encourages team members (at Superintelligent) to skip lengthy verbal feature proposals and instead “vibe code” new ideas rapidly—using code and AI to prototype before sharing.
  • “At this point we have a soft ban on describing new feature ideas. You just have to vibe code it up.” [40:45]

10. Naming: AI Weakness

Timestamps: [42:34–44:08]

• LLMs are consistently bad at creative naming. NLW relays ongoing disappointment at cringeworthy or unusable suggestions.
  • “I've never had an LLM come up with a name for anything...that I actually thought was even remotely usable.” [43:43]

11. Automations & Agentic Workflows (Aspirational)

Timestamps: [44:09–46:11]

• Despite extensive AI usage, NLW openly admits to lagging in automated content handling—has yet to fully deploy automations through Lindy, Nan, or Zapier.
  • “You have my full permission to...castigate me if in six months I don’t have some automated...setup.” [45:07]
  • “Even for someone who spends all their time on this, there is a hurdle and a barrier...” [45:33]

12. Experimenting with Gemini & Other Models

Timestamps: [46:12–48:02]

• While primarily OpenAI-focused, NLW notes growing maturity and integration of Google Gemini in Docs/Sheets for non-AI work.
  • “Gemini just keeps getting better. The models are improving, the interfaces are improving, the tools like AI Studio are improving...” [46:31]
• Concludes that model landscape is changing; expects broader model usage in the future.

Notable Quotes & Memorable Moments

• On Accepting AI’s Flaws:
  “Honestly, it’s a testament to how good some of these editing features are that I put up with [Descript] dropping an entire episode every couple of months...” [07:04]

• On Forming AI Writing Habits:
  “The best titles are ones that find a way to be a single, clear, determined statement, not have multiple sections.” [13:33]

• On Custom Writing Styles:
  “You can create custom writing styles… These are akin to something like a midjourney style template, but for writing instead of for images.” [17:45]

• On Pitch Collaboration:
  “You have to assume some level of sycophancy is going to be the norm on your first attempt...” [33:28]

• On Apple’s Voice Recognition:
  “If you use an iPhone… you'll note just how unfathomably, offensively bad its voice recognition software is at this point.” [34:18]

• On AI Naming:
  “Good Lord, are LLMs just absolutely, phenomenally, almost hilariously bad at naming things?” [43:14]

• On Falling Behind (automation):
  “You have my full permission to...castigate me if in six months I don’t have some automated...setup...” [45:07]

• On the Evolving AI Landscape:
  “For now though, that's the current status...But it'll be interesting to see in a few months where the balance of usage is between Gemini, Grok, Claude and OpenAI.” [47:25]

Models, Tools, and Use Cases: At-a-Glance

| Task / Workflow | Tool/Model(s) Used | Notable Details | |---------------------------|----------------------------------|------------------------------------------------------| | Voice agent audits | Custom, OpenAI API | Strategic client insights | | Podcast editing | Descript (Underlord), VEED | AI-powered, time-saving features | | Short clip generation | Opus, etc. (not yet used) | Aspirational, for social video | | Writing show notes | ChatGPT 4o/4.5 | Functional, but avoids AI-titles | | Long-form copy & styles | Claude (Opus 4, latest) | Uses custom style prompts | | Research (deep) | O3 deep research, ChatGPT | For backgrounders, strategic preps | | Research (quick) | Perplexity | Daily factoids, numbers | | Strategy prep/ideation | O3, Grok4 | Strategic “collaborator” for ideas/outlines | | Voice transcription | Flow / WhisperFlow (WISPR) | On-device speech recognition | | Cover and asset design | Ideogram, Midjourney | Ideogram for detail/text, Midjourney for creativity | | Coding/prototyping | In-house tools, LLMs | “Vibe coding” to demo product/feature ideas quickly | | Naming | GPT/Claude/others (none satisfy) | Consistently disappointing results | | Automations | Lindy, Nan, Zapier (not yet) | No full pipeline in production yet | | Other LLMs | Gemini, Grok | Testing and expanding scope |

Final Thoughts

NLW’s episode is both a roadmap and candid reflection for anyone using or planning to integrate AI into creative and operational workflows. He’s transparent about the strengths and limits of current tools and models—and honest about simply not having it all figured out. The blend of practical insights, workflow tips, and a realistic look at what’s still hard makes this episode especially valuable for executives, content creators, and anyone trying to keep pace with AI’s rapid advance.