Best AI for copywriting, video, automation & more

Guten tag lovely people,

There hasn't been anything too major this week to update you on.

We've had back-to-back weekly drops of Sora 2, ChatGPT apps, and Claude Skills, so that should be enough to keep you tinkerers out of trouble.

I thought this would be a good opportunity, especially with lots of new hungry builders joining the newsletter, to give you the current lay of the land…

What tools are in my core stack right now, and what LLMs you should be using for what operations.

This is a longer one, but worth the read (or skim).

Different models serve different purposes. The real skill isn't knowing which LLM is "best", it's knowing which one to use for each specific task.

Here's my current top 6, from #6 to #1:

I haven't used DeepSeek much yet, but I'm paying close attention to what it represents.

DeepSeek is the most efficient LLM available. It delivers strong performance at a fraction of the cost of other models. And the fact that it came out of China, built with sub par GPUs and essentially caught up to the US, which had a massive head start, should get everyone's attention.

It's particularly strong for technical and engineering applications, which makes it valuable for automations at scale where cost efficiency matters.

It's weak at creative or narrative tasks. If you need writing, storytelling, or brainstorming, this isn't the tool.

Use it for: Running automations at scale to reduce costs.

I'm not using Grok much right now. I use Grok when I want unbiased answers. Most other models tiptoe around touchy topics. Grok doesn't.

Grok is connected directly to live X data. That means it has access to real-time trends, social listening, and cultural signals that other models simply don't have. For anyone building a brand, tracking public sentiment, or staying on top of what's happening in real-time, that's incredibly valuable.

I also wouldn't bet against Elon Musk. He's proven time and time again that he can turn ambitious projects into category defining products.

Right now, Grok is personality driven and entertaining to use, but it's still maturing. It's not the right fit for technical tasks or anything that requires precision.

If you live in Google Docs, Sheets, and Drive, Gemini integrates seamlessly. It's also strong for SEO and technical writing, which makes sense given Google's infrastructure and search dominance.

Gemini excels at advanced reasoning and accuracy on complex tasks like math and coding, and it supports multimodal inputs; text, audio, images, video (can analyse them well). I'm just not using it much. Gemini's creative output feels robotic.

Use it for: Google Workspace integration, SEO research, technical documentation, image/ video, audio analysing.

Perplexity is the king of research.

It's the single most accurate LLM available right now. And I'm pretty obsessed with it.

When I need to research something quickly and I need the answer to be right, I use Perplexity. It cites sources transparently, which means I can verify the information without going down a rabbit hole of fact-checking. Perfect for market research, competitive analysis, or any situation where you're making decisions based on data.

Their Comet browser with agents is also very neat.

Downside is limited creative generation. If you need to brainstorm ideas or write something from scratch, it's not the right tool. It's built for accuracy and research, not creativity.

Use it for: Market research, competitive analysis, fact-checking, sourced answers.

I use ChatGPT about 100 times a day. It's still not quite as much as #1, but it's close.

ChatGPT is the most versatile and user friendly AI available right now. It's the tool I default to for quick research, brainstorming, strategising, automation, and rapid fire questions: "How do I connect this to that?" "How many calories in a banana?"

The interface is clean, the responses are fast, and it integrates seamlessly into workflows I've already built.

ChatGPT can hallucinate on niche topics, especially if you're working in a specialised area. You have to fact check everything. And its writing is awful, in my opinion it's very easy to spot content written by ChatGPT.

Use it for: Quick research, brainstorming, strategy, automation, everyday questions.

I'm very bullish on Claude right now. I've moved most of my work into Claude.

Their Projects feature was already great, but after adding Skills (SOPs for AI), Claude is literally insane. They're also winning on vibes.

All my “AI employees” live in Claude as projects - with access to all my tools like notion, Airtable, gmail, etc. via MCP connections. And now they also have the skills that I want to give them like branding, social content writing, etc.

All written my written content; social posts, landing pages, video scripts, anything that requires nuance and depth goes through Claude.

Their model Opus 4.1 is exceptional. But it's so expensive, and you'll hit your daily limit in three messages. so I don’t really use it. sucks.

Sonnet 4.5 is also insane. This is what I use mainly.

If I'm doing serious work, it's in Claude.

n8n takes the crown for automations. It has a steeper learning curve, but there are hundreds of free courses on YouTube and thousands of pre-built templates online for almost any general automation you could need; onboarding, content creation, research, etc.

It’s also very easy to find cheap ($15-25/ hour) n8n specialists on upwork.

n8n is best for: complex, multi-step automations that need to be reliable and scalable.

Lindy is a solid option for non-technical beginners. it's more like Zapier on steroids. I haven't played around with it long enough to confidently tell you to drop everything and use it, but Lindy is much better for quick, surface-level automations (and browser-use automations). n8n is the most robust.

Think Figma vs Canva.

Lindy is good for: Quick automations, browser-based tasks, lightweight workflows.

Airtable is the database for all my automations to run off. It holds all my data. think Excel spreadsheets built for automations to run on.

For example, if I want to automate social media content creation, the n8n workflow needs somewhere to output the content it's created:

ElevenLabs takes the cake for anything voice-related.

Voice cloning, text-to-speech, multilingual dubbing. if it involves audio, ElevenLabs does it better than anything else available right now.

I use it for narration, video voiceovers, AI voices for ads and turning written content into audio formats.

Notebook LM is in my core stack to learn everything 10x faster.

Each "notebook" is a subject or topic you want to learn about. You upload your source material; courses, YouTube videos, documents, etc. and now it's a mini brain that's an expert on that topic.

I was genuinely in disbelief when I first used it. You can generate quizzes, flashcards, docs, blog posts, reports, explainer videos, and podcasts.

I recorded a short demo a few weeks back. You should watch it here (guaranteed value)

Use it for: Accelerated learning, creating study materials, turning raw content into structured knowledge.

Notion for project management and docs.

The UI is great. It’s simple and effective. And the Claude/Notion connection is great. I can chat with Claude and ask it to reference different notion pages for context, and get Claude to create/ update my Notion pages.

I know Notion is pushing heavy on the AI aspects of their software, but tbh I haven't even used it once. Notion is my brain and hub for everything I do; docs, project management, tracking, organising. It's where everything lives.

I used to use Gamma for slide decks and presentations, but since Claude got the Skills feature last week, I just use Claude's PowerPoint skill and export it into Canva for final touches. More on this next week and on my YouTube.

I recently started using Wispr Flow for voice-to-text.

Words can’t begin to describe how useful this has been. Saves me genuinely hours each week. I use this most when chatting with AI; it's just speech-to-text, so I don't have to type anything. pls use my referral link :)

Use an aggregator. Makes everything so easy as AI moves so fast. An aggregator is just a site that allows you to access all the different AI video and image models in one place with only one subscription.

I like OpenArt and Higgsfield. I've heard Artlist is good too.

Best video models right now:

Both Sora and Veo generate with audio (sound effects, dialougue, music, etc).

Best for photo:

For most of my website needs, I've been using Framer (not a vibe coding tool I know).

Took me a few days to learn with no technical or design background.

It's perfect for quick, easy shipping of websites and landing pages with beautifully designed templates. Just change the copy, and you're good to go.

I haven't spent as much time mastering vibe coding yet, so my breakdown here is a little lacklustre, but I'm confident I can point you in the right direction:

Lovable: Great place to start as a complete beginner for both apps and websites. I've spent a lot of time with Lovable building landing pages. Volo Builds has done an incredible demo showcasing what's possible.

Claude Code: Takes the cake for both websites and apps, but it's a very technical tool with a steep learning curve. If your goal is to go beyond cookie cutter apps and actually understand what's happening under the hood, Claude Code is the one that will level you up fastest.

Greg Isenberg just did a tier list video ranking all the best vibe coding tools on the market. I would be stupid to not just direct you there. these guys know more than me about coding and apps.

I plan to spend the next couple of weeks learning and mastering vibe coding for websites and landing pages.

So I'll report back to you soon 🫡

In other news, I finished up a 72 hour fast yesterday (just water). I like to do one every year for the health benefits + the mental challenge.

We’ve become so indifferent about food, and doing a prolonged fast gives you this newfound sense of gratitude.

That’s all folks. Go have fun with your new toolbox!