Your prompts are leaving out 80% of what you're thinking.
When you type a prompt, you summarize. When you speak one, you explain. Wispr Flow captures your full reasoning — constraints, edge cases, examples, tone — and turns it into clean, structured text you paste into ChatGPT, Claude, or any AI tool. The difference shows up immediately. More context in, fewer follow-ups out.
89% of messages sent with zero edits. Used by teams at OpenAI, Vercel, and Clay. Try Wispr Flow free — works on Mac, Windows, and iPhone.
Hi Guys!!
This week we cover:
GrokBot blew me away in 5 min - Here’s why
ChatGPT recommended a fake product - What you need to know
Experiment I tried - Can Claude/ChatGPT/Grok edit my raw videos + my honest thoughts
GrokBot was Shocking
I wasn’t going to try GrokBot as I’ve never really stayed much longer than a day or 2 before forgetting about it.
But this time, it was REALLY impressive… It’s like Buzz/Slack/OpenClaw/Hermes. I know I spoke about Buzz last week and yes I thought it was that too but the output was unimpressive as I mentioned (I also spent a few hours on testing).
GrokBot on the otherhand I had set up AND tested in 5min. Here’s the video I made straight after.

I was super impressed especially for these reasons:
Easiest + fastest set up for non technical people (just click buttons + login)
Understands context WAY better than any other AI I tried out of the box (I only connected gmail, instagram)
Proactive (did things without me being super specific with prompts)
Using latest Grok 4.6 which is comparable to GPT 5.6 Sol & Fable 5 at a fraction of the cost
Works on its own virtual cloud computer (so you don’t need to keep your laptop open)
Create routines + teach it how to do your work by recording it
Improves as you use it more - without you telling it to remember
Feels like messaging your AI team mates in Slack (connect a few bots in 1 chat so they work together)
iOs mobile + desktop available
Example 1: Content Ideas based on my Instagram account
I ONLY gave it a link to my instagram account. It proactively went to research me and made 6 ideas that matched my content AND based off current news (obviously since it’s Grok it has access to X threads)
Codex and Claude can do this too but it was much more proactive and less prompting out of the box

Also made me a Brand Nat Content Pack for this week - without me asking!

Example 2: Inbox Triage
Again, I only gave access to my Gmail. Grokbot ranked the top emails I should focus on (short reason why, what is needed of me), who I should not answer at all (low ball offers).
I can’t tell you how long I spent trying to drum this into Claude + Codex + Hermes with very specific instructions multiple times over many weeks! Grokbot went straight in to draft the email and also noticed how I normally write emails so it imidated this - without me asking!
Example 3: Uses tools pretty well
I uploaded my raw video and told it to copy another creator’s editing style (just sent an Instagram link).
Out of the box it gave me probably the closest replica compared to Codex/ClaudeIt was able to grab screenshots of news articles, add motion graphics etc. Now in the end I chose Claude’s output BUT to be totally fair, Claude had a 20 page doc uploaded with instructions and alot of back and forth. Grok had nothing but a simple 1 sentence prompt with a few follow ups.

Downsides
Only uses Grok model
Price point of $200/mo (via Cursor Ultra) or $300/mo on SuperGrok Heavy which may deter many
Not opensourced like Hermes, Openclaw and Buzz
POLL/SURVEY
ChatGPT Recommended a Fake
This might be one of the most important AI shopping experiments I have seen.
A marketer created a fake natural deodorant brand called “Morrowen”. There was no product to buy, no customer history and no clinical testing.
She bought the domain for $11.25, created a simple website with three pages of copy, published one Substack article, then left it alone.
Three weeks later, ChatGPT with web browsing started recommending it 😱

Now, the experiment was much narrower than the viral headline suggests:
Morrowen appeared for 2 very specific questions about magnesium deodorant for sensitive skin and irritation from baking soda
By day 30, she reported it appeared first in 4 out of 4 answers to those narrow questions
However, it did not appear for the other 13 general shopping questions
Claude and Gemini did not mention it though
So this is not proof that anyone can dominate every AI recommendation with an $11 website.
But here is the worrying part.
In one answer, ChatGPT placed the fake brand beside MAGS Skin, a real product listed by the National Eczema Association. MAGS Skin had completed third-party safety testing on 60 people with sensitive skin.
Yet ChatGPT presented both brands in the same confident voice and format. One had independent evidence. The other had persuasive website copy for a product that did NOT even exist.
My take: AI is creating a new shopping shelf, but its confident answer may come from limited research.
A study of 4,706 searches found GPT-4o Search showed about four sources in a typical answer. Your wording also shapes what ChatGPT searches.
Treat AI answers as a useful starting point, not proof that the whole internet has been searched or that every source supports the full conclusion.
TREAT TO TRY - Claude Edits My Videos
Can Claude edit your raw videos from start to finish?
I saw this video about using Claude code to edit your raw videos and saving 4-5 hours. Does it actually work? I tested it myself so you don’t have to.
In fact I tested it out in both Claude, ChatGPT and Grok!
If you want to see my video comparison, here they are.
For the below photo reference:
1 = Claude Code (Fable 5)
2 = ChatGPT (Codex/Work with GPT 5.6 Sol Med)
3 = GrokBot (Grok 4.6)
I basically prompted Claude to install this resource, uploaded my raw video and gave it a reference video (a style I wanted it to base it off).
From there, just talk to it like a human to make changes. Everyone will have their own taste so I will leave that up to you.

Simple prompt to give Claude
Hook
Grok (3) stuck to the original video the best (blue font with emoticons)
Claude (1) and ChatGPT (3) both took on Claude colours since the video was about Claude. I do appreciate the bigger text which is more pleasing on the eyes

Dynamic Animations/ Motion Graphics
Claude (1) and ChatGPT (2) did quite well with animations which fit what I verbally said but I think ChatGPT wins slightly due to the ease of reading and more polished look. Claude was simple but often too small and empty
Grok (3) didn’t do as many but to be fair, I didn’t prompt it nearly as much as the other two AIs

Tool use
All 3 grabbed screenshots of up to date articles but Claude (1) did a better job with clean polished editing of the article headlines

In my opinion, it was Claude or ChatGPT that took the cake. Initially I thought this wouldn’t be good enough to use but I ended up posting Claude on my socials just to test.
To my surprise, it actually did better than my normal posts + genuine discussion. It was one of the easiest posts I’ve done plus I would have spent hours on this!
Which one do you think did better?
If you want to try it yourself, I’ve made a free resource pack you can download here.
Just upload into Claude code and use the prompt below to install:
Read README.md and install this video-editing resource pack for me. Complete the setup, check that everything works, and tell me when it is ready.
(I made sure Claude tested this resource again to make sure it would pass the video test! Hope you enjoy!)

That’s it for this week! 😊
Cheers,
Nat



