just your friendly neighbourhood printer fixer 100x.iamdevloper.io - 100x engineer exam 📮 the friday email: iamdevloper.io/newsletter
UK
- Tweets 9,425
- Following 0
- Followers 393,660
- Likes 1
I finally reached the magic number of $5M in net worth 🥳
This isn’t all cash. It’s everything combined:
+ My 36 startups (90%+ of their value comes from DataFast & TrustMRR)
+ Investments (85%+ in the S&P 500)
+ Cash
What surprises me most is getting here at 33 without ever:
- Working more than 8 hours a day
- Building anything grandiose
- Working for someone else
- Doing work I don’t like
- Going to an office
Don’t you dare give up :)
6
12,803
claude when it reads a secret key from your codebase into the chat context
buddy, I do not care
17
12
232
15,605
Me explaining to Claude that it needs to scan our entire git history for the past week and summarise for me because I have standup in 5 minutes
3
14
344
17,286
Which one are you?
Everyone thinks they are a 10x engineer.
Forty questions will tell you what you actually are.
100x.iamdevloper.io
6
5
26
20,104
I'm impressed...
Everyone thinks they are a 10x engineer.
Forty questions will tell you what you actually are.
100x.iamdevloper.io
10
15,025
A candidate has been assessed at 1000x.
Forty questions. Forty correct. Five minutes and forty-eight seconds.
The Institute calculates the probability of a candidate of average ability achieving this at 1 in 5,496,666,084,465.
No further comment will be made.
100x.iamdevloper.io
with 183 people completing the exam (100x.iamdevloper.io/), one person reached 632x engineer status
getting close to 100x.iamdevloper.io/hall
1
18
17,576
with 183 people completing the exam (100x.iamdevloper.io/), one person reached 632x engineer status
getting close to 100x.iamdevloper.io/hall
2
1
11
34,412
claude considering all the possible options before settling on creating a 300-line SUMMARY.md
3
1
76
14,464
being asked to design a scalable system during a tech interview whilst the company's status page shows outages
1
7
71
14,335
github.com/benhowdle89/matin…
Demo:
matinee.pages.dev/
Staged cursor performances for React. A programmable ghost cursor that moves, clicks, types and scrolls like a human, then exports itself as an animated SVG.
1
4
52
13,459
Speak to my Agent (not me)
Dark - iamdevloper.creator-spring.c…
Light - iamdevloper.creator-spring.c…
Cap - iamdevloper.creator-spring.c…
2
25
13,164
Speak to my Agent (not me)
Dark - iamdevloper.creator-spring.c…
Light - iamdevloper.creator-spring.c…
Cap - iamdevloper.creator-spring.c…
1
2
10
31,154
Claude: it's getting late, you've shipped so much today, maybe it's time for you to get some rest
Me: maybe it's time for *you* to get some rest Claude
Us:
24
12,490
"That’s a genuinely great question — and honestly, it’s not really about the em dashes, it’s about what the em dashes quietly re-"
1
24
13,291
when claude gives me a 2-3 week timeline, but I tell it it could do it in 30 minutes, and it does it in 7 minutes
5
71
9,261
now that all of our jobs look like Locke diligently pushing the buttons in the hatch/saying "yes" to Claude.....what are we supposed to do now?
3
1
16
6,361
> ok, give him a deadline to use Fable by. Ok, now extend it once he's used all his token
3
43
8,042
> hold on. your disaster recovery plan is ‘it’s probably still in my Claude chat history’?
1
2
101
16,401
Bloome keeps everyone's input, humans and agents, in one shared conversation, and when someone joins late an agent just summarises what they missed.
The agents don’t just argue, they refine the work step by step until it becomes a usable final output you can actually act on.
asynchronous collaboration where nobody has to say "quick recap for those just joining."
1
1
3,009
also every model's in the same window now. Claude, ChatGPT, DeepSeek in one thread instead of me copy-pasting between four tabs like it's 2023.
the tab graveyard is healing.
1
1
171
you're the editor, not the typist. genuinely a better shape for real work than asking one model and hoping.
1
804
that’s what makes it feel different when you actually use it: aisecret.co/Bloome
the bit that sold me: I asked for market research, the first agent confidently wrote a market-size number, the second agent went "source?" and it didn't have one.
single-model chat never does that. it just agrees with itself and you find out later.
1
1
2,240
Bloome lets multiple AI agents into one chat where they check each other's work. one drafts, one disagrees, one finds the thing both missed. finally, a team meeting where nobody's clashing.
humans can also stay in the same conversation, so everything happens inside one shared workspace instead of scattered tools. 🧵
18
19
41
65,429
after receiving that Vercel bill
Was thinking today how nice it is to process 1.5 billion requests per month and @vercel "just handles" the scale for us. They've really built something special that allows eng teams to focus on building.
1
2
52
11,665
"I don't want a Creative Cloud subscription. I just want to pay for Photoshop."
82
340
4,479
109,788
that 3s gap after hitting Leave in a Zoom call and it asking if you're sure you want to leave the call but you've already said goodbye to everyone
1
1
10
5,314
"Mark the report as urgent. Make them work longer hours to complete it. Ok, now make sure no stakeholder looks at it for at least 3 days after it's been submitted."
2
55
7,407
sat on the edge of the hotel bed for 30 minutes after a day of meetings, before needing to head back out for the team dinner
1
13
4,019
There's a new kind of coding I call "vibe coding", where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It's possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper so I barely even touch the keyboard. I ask for the dumbest things like "decrease the padding on the sidebar by half" because I'm too lazy to find it. I "Accept All" always, I don't read the diffs anymore. When I get error messages I just copy paste them in with no comment, usually that fixes it. The code grows beyond my usual comprehension, I'd have to really read through it for a while. Sometimes the LLMs can't fix a bug so I just work around it or ask for random changes until it goes away. It's not too bad for throwaway weekend projects, but still quite amusing. I'm building a project or webapp, but it's not really coding - I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.
5
26
8,146
when my colleague's email reply starts with "Excellent, now that I have the full picture..."
1
4
77
10,454
mfw I get an email telling me claude mythos found a bug in some code I wrote 30 years ago
1
5
34
6,728
also uses 70% fewer tokens than OpenClaw
which matters when half your token budget was just the agent figuring out what to do next
1
1
2,833
it runs in a sandboxed architecture with encrypted storage, role-based access control, and audit trails so unlike some other agents i could mention, it does not hand your credentials to whoever asks nicely
1
1
670
this is @Pokee_AI PokeeClaw
an AI agent that runs real workflows across your apps - with control, context, and continuity
2
2
1,041
I typed one prompt to read my inbox, screen AI event invites, handle the responses,
and prepare everything needed to attend.
I did not touch a single email.
Here is what happened:
✔️ Scanned my inbox, picked the best 3, and replied to every invite.
✔️ Sent me a full summary on Slack
✔️ Prepared tailored talks and IG posts
8
25
48
47,325
told UI Bakery "build me an internal Twitter analytics tool"
two minutes later: sidebar nav, KPI cards, follower growth chart, engagement trends
I did not type a single closing bracket
1
2
951
anytime I feel bad about myself, I think about this interaction and immediately feel better
1
54
21,961
benhowdle.im/closeloop/
me: I need to close my open loops to prevent burnout
also me: *adds 'close open loops' as a new open loop*
1
2
5,007
day 1: "I just need a simple backend"
day 2: setting up Postgres
day 3: writing Express routes
day 4: auth middleware
day 5: deploying to AWS
day 6: debugging CORS
day 7: logging and monitoring
day 8: still no frontend
there's gotta be a better way to do this
reqres.in
11
6
48
24,706
I just...don't believe the $46k hosting costs
The Epstein email website racked up a $46,485.99 Vercel bill
All the replies were:
1. Use Cloudflare
2. I could host this on a VPS for a ham sandwich
I dug into the site to see what could be causing this and if it would be cheaper or even possible with 450 million page views
piped.video/watch?v=ihnGot4n…
4
2
19
10,271
> you know, you have to actually market your app now, and not just build another idea
me:
1
1
5
4,340
Higgsfield put Claude’s reasoning engine inside a motion design system.
Not pattern-matching.
Not vibes-only.
Actual intent → motion logic → controllable output.
Same Claude that’s good at code. Now doing motion.
1
2
1,733
It also runs tasks in parallel across environments and learns your codebase as it goes.
No model picker. No micromanaging.
Reasoning, coding, UI, testing all routed automatically.
You can submit multiple tasks, close your laptop, and come back later.
That quietly changes how developers think about time and focus.
1
2
262
It is called Quest 1.0 by Qoder.
This is not a copilot and not “watch me code”.
You define intent.
It generates a spec.
You approve what “done” means.
Then it executes autonomously until it works.
1
5
2,462
I have been testing something that actually feels different.
Instead of prompting line by line, I described a goal, reviewed a spec, approved it, and walked away.
Came back later to working, verified code.
That distinction matters.
1
3
2,682
