Product Manager Morty

cute ngl.

@sflorimm is all of them a bad answer?

be honest devs, how many side projects are still unfinished?

@thenowhereway fell in love with the new possibilities not the coding part.

A lot of people fell in love with coding after AI arrived. Would that love survive without it?

@mandylu pain.

daily reminder

how many people did you talk to were neurodivergent / severe case of functional ADHD. I think the number can be higher but yes, humans can never multi thread as efficiently as machines. The more abstraction the more chance non-determinism comes up and bites you. Being able to think and work in meta layers is really helpful but i think you're conflating two very different challenges.

Humans can only manage 5 AI agents max I asked everyone I know: "How many agents can you manage simultaneously?" The consensus: 3 is hard. 5 is the absolute limit. Nobody can effectively manage more than 5 agents at once. Managing 3-5 agents turns you into a context-switching nightmare. Channel A, Channel B, Channel C. Your brain becomes a pinball machine. Two key insights: 1. The future isn't humans managing agents - it's agents managing agents. I can't personally manage 100 agents. My brain would explode. The only path to scale is having a meta-agent manage my agent workforce. If I only manage 5 agents, I'm basically a small team lead. M0 level at Facebook - managing 5 direct reports. But if I can manage 50 agents through AI management layers? That's a completely different power level. 2. The bottleneck is task duration, not task complexity. If an agent bothers me every minute, I can only handle 1 agent. If it's every 5 minutes, maybe 3 agents. If it's every 10 minutes, possibly 5 agents. The breakthrough everyone talks about - "long horizon tasks" - isn't just about AI doing complex work. It's about AI working independently long enough that humans can actually parallel multiple agents. Real-world implication: Facebook now ranks engineers by token usage to measure AI adoption. But you can't burn serious tokens by manually managing agents one-by-one. To hit the top of that leaderboard, you NEED agents managing other agents. That's the only way to achieve massive token consumption. The human cognitive limit is real. 5 agents maximum. Everything beyond that requires AI management layers. We're exploring this at our company. I think "Agent Manager" as a product category will emerge very soon. The question isn't "How good are you with AI?" It's "How many management layers can you orchestrate?" That's where the real leverage lives.

The run was heavy: 375 agents, ~19.5M tokens, ~3.3 hrs, ~25 verify agents stalled/failed. It delivered, but it's not a cheap thing to re-run casually. workflows is like /goals + cronjobs / daemons on steroids... the update(s) are now increasingly less about the model and more about creating harnessing features that help people take advantage of the model. are we as a society even ready for Opus 5.0 ?

@mandylu pyramids,,

we still have no satisfying theory for why AI works

@natxwang says hi to their colleague on monday morning with "yoooo -- did u see _____"

How to spot a builder on X: 1. Their bio says "building" something 2. Has bought 18 domains for ideas that don't exist yet 3. Every problem becomes a startup idea 4. Tweets "shipping" instead of "working" 5. Has a mailing list before a product 6. Gallery consists of random screenshots of other apps and sites 7. Owns more SaaS subscriptions than customers 8. "Just vibecoded this in 3 hours" 9. Has rebuilt the landing page 14 times 10. Measures success in GitHub commits and coffee consumed 11. Replies "interesting..." whenever they get a new idea 12. They have a digital brain as a backup 13. Every conversation ends with "actually, I'm building something for that" 14. Has 45 tabs open and needs every single one 15. Thinks weekends are just weekdays without meetings 16. Hasn't touched grass but has touched 14 AI models this week 17. Every post mentions "AI" What is missing from this list?

@dev_gen88926 @suryanox7 the learning that comes from automating a task that doesnt need automation may come in handy later down the track. learning on someone elses dime is super non-linear

@dev_gen88926 @suryanox7 this

software engineers will automate a 10 minute task using a script that takes 6 hours to write and still feel like they won

@omgsidewalks ur asking the wrong part of the internet this question ///

Do you guys actually enjoy being home alone all day without seeing or interacting with anyone ??

An RTX 3090 does ~35 trillion calculations per second. To match it by hand, you'd need ~4,400 Earths full of people, each solving one math problem a second, every second.

@abdo52085206 @yanliudesign it was only a short time ago that claude design was a seperate stream of tokens. it was fun whilst it lasted.

@yanliudesign very nice stuff, I would guess that burns too much tokens right?

A quick recap of what I experimented with Claude Design in May: ✧ Use a mood board to define the look & feel ✧ Add animation + interaction to level up the design ✧ Generate an image of my vision, turn it into an interactive prototype ✧ Claude design is great at graphic design More to explore in June ｡･:*:･ﾟ★,｡･:*:･ﾟ☆

you can still have all of those things -- the difference is that a lot more of those things now happen internally in your head. that doesn't mean it's completely lost -- its entirely possible to have two completely diverging / opposite opinions on the same idea and work towards convergence by yourself (+AI)

AI is taking the fun out of the build and I think we need to say that out loud. Building was never just about the output. It was about the people in the room, the late nights, the arguments, the breakthroughs you only reach *together*. That friction? That's where creativity lives. Now we build alone and in silence. Have we traded collaboration for convenience? I'm not sure we got the better deal. 🤔

@natxwang im down to join

AI is fun. But tinkering alone all day? A little less fun 😅 Started a casual Discord for AI tinkerers + builders to: → talk about the latest updates → brainstorm ideas → get feedback → share workflows + automations There's about 20 of us now. A mix of founders, builders and people who are curious and just getting started. Mostly just a good space to vibe with others equally obsessed with AI. The plan is to keep the community small allow comfortable and open conversations that contribute. Lurkers and sus individuals will be removed. Leave a comment if you want in. Trying to avoid bots and minimize scammers.. Make sure your DMs are open!

things i find severely lacking in current llm benchmarks but i feel are important for iterative, open-ended knowledge work: - problem-setting. a major part of creative/exploratory work is defining what the problem even looks like. benchmarks skip that and hand you the problem pre-formed. it's the same issue with using exams to predict how someone handles real world problems.

@wordgrammer @nickcammarata thinks and works is increasingly blurring

@nickcammarata A middle ground in between the two is someone who works 60 hours a week but thinks about work 120 hours a week. I think Jensen fits this description? From interviews at least

Working 100 hours a week is characteristic of jobs where you “work” very little, but need to always be on call (think: investment bankers). Fields that require deep creative work or technical precision usually get 4 productive hours each day. Agents moved SWE from (2) to (1)

@_jzhao style is styling

i then proceeded to stay up very late and redo my personal website again (experiments with procedurally generated dappled light and dithering!)

experiment with a memory system that keeps rewriting itself: a hopfield network remembers an alphabet. as memories decay, it begins to hallucinate glyphs it was never taught - forgetting becomes a way of inventing.

@helloiamleonie providing visibility on where the jaggedness is

just curious: what’s the most useful thing your OpenClaw, Hermes Agent, etc. is doing for you?