Michael Johnson

7K posts

Michael Johnson banner
Michael Johnson

Michael Johnson

@onemoremichael

Dad, Co-Founder, Context Architect, Buchholz High School, Best Ideas Award winner (3rd grade, Norton Elementary)

Katılım Temmuz 2018
7.5K Takip Edilen660 Takipçiler
Michael Johnson
Michael Johnson@onemoremichael·
@sama It still feels absurd that for $200 I get nation state level SOTA intelligence. Pretty sure that is not gonna last long and access/ability to deploy capital at a problem will far out pace tooling like GPT-5.4-Pro (we may already be there narrowly) but for now it’s kinda amazing
English
0
0
0
29
Michael Johnson
Michael Johnson@onemoremichael·
Being early is quite an advantage! My parents dear friend and home builder from back in the day built himself custom software to manage his builds. He showed it to some Dr friends and they thought it could work for them, thus Medical Manager was born. Lots of value was created! I’d be willing to bet no VC was around but I’ll have to ask him
English
0
0
0
18
Steph from OpenVC
Steph from OpenVC@StephNass·
Chess․com Revenue: $100M+ VC: $0 Mailchimp Revenue: $700M+ VC: $0 Zoho Revenue: $1B+ VC: $0 Midjourney (via @onemoremichael) Revenue: $200M+ VC: $0 Butcherbox (via @DanielGulati) Revenue: $600M VC: $0 Dyson (via @PatrickOCR) Revenue: $8B+ VC: $0 Who else?
English
47
31
481
47.5K
Michael Johnson
Michael Johnson@onemoremichael·
That’s a really good one! Especially because it was in 79 right when ERISA changed and let pensions get in the game and VC as we now know it started to emerge. If I were writing a proper essay on this I would 100 percent start there as the timing is just 🧑‍🍳😘 perfect for a starting point.
English
0
0
1
70
Nous Research
Nous Research@NousResearch·
Hermes Agent wrote a novel. "The Second Son of the House of Bells" runs 79,456 words across 19 chapters. The agent built its own pipeline to do it, using the ame modify-evaluate-keep/discard loop as @karpathy's Autoresearch but applied to fiction: world-building, chapter drafting, adversarial editing, Opus review loops, LaTeX typesetting, cover art, audiobook generation, and landing page setup. Book: nousresearch.com/bells Code: github.com/NousResearch/a…
Nous Research tweet media
emozilla@theemozilla

it's been a longstanding dream of mine build an ai system that can tell a compelling story. it's what got me started in the space in the beginning, and with Hermes Agent I finally pulled it off 100% written, typeset, etc. by Hermes Agent those at our gtc event got hard copies🤗

English
64
94
1.2K
153.1K
Leg Before Teebo
Leg Before Teebo@dmesg·
@MParakhin This may be a stupid question but why not just use the computer you already have?
English
2
0
1
220
Michael Johnson
Michael Johnson@onemoremichael·
@bcherny Does it insert/impact the in-progress run? If no, once the in-progress run is finished does it have awareness of the /btw, and does the /btw branch have awareness of the main output?
English
0
0
3
414
Michael Johnson
Michael Johnson@onemoremichael·
@lukOlejnik So many orgs laid off engineers who were their most diligent coders and slow to adopt AI, when those are the key skills for effective review. That and managers, which is essentially the other part of the job now 🤦🏻‍♂️
English
1
4
61
8.1K
Lukasz Olejnik
Lukasz Olejnik@lukOlejnik·
Amazon is holding a mandatory meeting about AI breaking its systems. The official framing is "part of normal business." The briefing note describes a trend of incidents with "high blast radius" caused by "Gen-AI assisted changes" for which "best practices and safeguards are not yet fully established." Translation to human language: we gave AI to engineers and things keep breaking? The response for now? Junior and mid-level engineers can no longer push AI-assisted code without a senior signing off. AWS spent 13 hours recovering after its own AI coding tool, asked to make some changes, decided instead to delete and recreate the environment (the software equivalent of fixing a leaky tap by knocking down the wall). Amazon called that an "extremely limited event" (the affected tool served customers in mainland China).
Lukasz Olejnik tweet media
English
971
3.3K
19K
29.8M
Michael Johnson
Michael Johnson@onemoremichael·
You are not wrong, but I also have empathetic concern/dread for those that will/are struggling because they lack your intrinsic curiosity and agency. It could get rough out there for them. So I vacillate with the daily wonder around what I get done and then concern for those getting left behind. My answer to try and alleviate that dread is to teach as many people in my sphere as possible in hopes of igniting their curiosity and drive to participate in this step change. I think you are doing the same with your content 🫡
English
1
0
2
76
Jesse Genet
Jesse Genet@jessegenet·
Spicy take 🌶️ If this age of AI gives you existential dread instead of giving you wings you weren't that good at what you 'do' We all have superpowers and propellant to get work done better and faster, but there is nowhere to hide a lack of vision for 'what' to get done
English
53
26
286
15.3K
Michael Johnson
Michael Johnson@onemoremichael·
So worth it! We all need to constantly consider what context we have and the system does not. Formalizing that so the tooling can reason about what blind spots it may have and lean into capturing it is wildly valuable. I’ve been treating docs as a first class citizen in my Context Map system but need to put more effort into capturing my off system context capture. github.com/Onemoremichael…
English
0
0
1
472
Allie K. Miller
Allie K. Miller@alliekmiller·
My team just held a one-hour context engineering sprint. Massively valuable. We blocked off everyone’s calendar for one hour, opened up Claude code, connected our context vault prompts and interview style, shared best practices ahead of time, and dictated back-and-forth with Claude for an hour (while walking for some of us!) with Wispr Flow. Our file system is our agent brain. Taking an hour or two to massively improve that is well worth it.
English
27
13
166
36.3K
Michael Johnson
Michael Johnson@onemoremichael·
@Dishantpanc5 @kellylsims I agree completely. I think it is fully logical that the DoW not use a certain vendor because their products TOS are to restrictive. I think it is fully logical that a private company has TOS for how its products are used.
English
1
0
0
49
Dishant
Dishant@Dishantpanc5·
@onemoremichael @kellylsims They never said they wanted the autonomous weapons NOW, they are negotiating the agreement when they are allowed for use. This is clever way to hide the core issues that anthropic wants to have control on how models will be used. I read it both sides. It became very clear now.
English
1
0
1
146
Kelly Sims
Kelly Sims@kellylsims·
"What concerns me about Amodei and Anthropic in particular is the consistent pattern of being singularly focused on being the one winner with all of the power, with limited consideration of how everyone else may react to that situation." This is a thoughtful piece on all this.
Stratechery@stratechery

Anthropic and Alignment Anthropic is in a standoff with the Department of War; while the company's concerns are legitimate, it position is intolerable and misaligned with reality. stratechery.com/2026/anthropic…

English
17
8
175
29.9K
Nathan Lambert
Nathan Lambert@natolambert·
Every Anthropic employee proudly amplifying their company comms and 0 supporting Sama’s weird scooping up of the DoW contract is pretty telling.
English
85
128
4.1K
252K
Michael Johnson
Michael Johnson@onemoremichael·
@toddsaunders Bonus part is there is often a huge token cost from ingesting historical data to get up and running
English
0
0
0
333
Todd Saunders
Todd Saunders@toddsaunders·
So many founders are doing this now. My inbox is flooded with stories like this.
Todd Saunders tweet media
Todd Saunders@toddsaunders

I got a DM from a CEO yesterday who charges customers $5,000 for onboarding. Not as a fee.. but as a deposit. If the customer (ICP is SMB $1-5M in revenue) completes onboarding within 35 days, they get every dollar back. If they don't, the company keeps it and uses it to fund a dedicated onboarding specialist to finish the job for them. (don't worry they call the onboarding specialist a forward deployed engineer) It sounded a bit crazy at first, but the results are pretty interesting. Before the deposit, their average onboarding took 67 days with 10% churn during the process. After, it dropped to 31 days, and 3.5% churn. The deposit didn't just speed things up..... it filtered for customers who were serious about implementing, which is a leading indicator of long-term retention. This year so far, 98% of customers have completed onboarding within a 30 days. Is anyone else doing something like this with setup fees for onboarding?

English
19
16
667
263K

Keşfet

@sama @StephNass @DanielGulati @patrickocr @FrancisPSantora @NousResearch @JSupa15 @karpathy