TB

senior product manager (ic) + product designer

[Andrej Karpathy @karpathy]

One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10 minutes, total mess, anything goes, full stream of consciousness. Sometimes I declare it up top, something like "switching to speech recognition sorry for any typos...". Sometimes I turn it into a small interview of a few turns. But I find that the LLMs are somehow very good at reconstructing long incoherent rambles and often their echo of your own tangle of thoughts comes out quite a bit cleaner than what you started with. The result is that you improve the mind meld and have to correct things less from that point on.

[TB @tb8424]

last weekend I set out to build a chatbot as a learning experience, and i accidentally ended up building a publically available, searchable memory of all my content.

it includes X replies, Slack replies and everything product-related that I've put out

runs on DeepSeek through @OpenRouter , is traced via @langfuse, deployed via @vercel and pulls content from 2 @upstash db's

[TB @tb8424]

@Madisonkanna https://t.co/OTf2BEFP8N

Really enjoyed this one

[TB @tb8424]

personally i use it a bit differently (i think by writing) so the AI stuff is made out of a bunch of skills that review my work and:
- clear the typo's cause if i write fast i might make mistakes but i don't want to stop the momentum
- force me to take a stance (stops me from doing red-team blue-team debates to the point where the article doesn't really say anything) - your alpha angle
- takes a "i hate this post and this is why" so i can address areas where i'm vague in my explaining / try to sound too smart for my own good

i found that the AI's desire to please people is reflected (in the case of a fully outsourced article) in a weak pov that tries to please everyone

[TB @tb8424]

@RaminNasibov https://t.co/MAyCbBZjDJ

I still think about this after years of seeing it.

[TB @tb8424]

imo it creates breathing room for shapes and proportions to stand out. I liked the Braun hardware (not black, but still monochrome) because makes the engineering craft stand out

i am not against colours (they are a part of "feeling" as much as shape and proportions) but sometimes they might be used as cop-outs for boring shapes

same with cool clothing that is cut well, or cars (i did a small drawing class on car shapes and the guy said you can figure out if a car is unique if you can figure out what it is in 3-4 rough lines on a paper)

on why black in itself, for me it carries more "weight" as a monochrome choice versus white, i find myself being less tempted to add things to fill space

[sphinx @protosphinx]

Renault pls make this a production car and you’ll instantly become the coolest car maker I guarantee it

[Ivan Landabaso @IvanLandabaso]

Startup sales gold:

[TB @tb8424]

@trq212 I created a similar skill called /design-shotgun that helps me brainstorm different design patterns for a problem that I'm having and the output of that in HTML as well. It's cool to see that you can extend that to other similar use cases. thanks for this!

[TB @tb8424]

working less on the same tasks yes, but the number of tasks then increases to fill capacity

same way computers made us faster -> but workload increased

i don't think the productivity gains transfer to the user, but rather to the company the user is working for

with the exception of "hidden ai" use: people who use it in companies where it isn't the norm, and don't tell, so they do "work less"

[TB @tb8424]

https://t.co/5otg8T8ShY

[TB @tb8424]

@Prathkum not sure if it's just me normalising to it, but earlier models were "same style but a lot more powerful" but now i feel that it comes with a bit of a "will this have a diff vibe". Started feeling it around gpt 4 vs 3.5

[TB @tb8424]

i have a bit of a weird way of thinking about this

the closer it feels to you waving a magic wand the better it is

Eg: you have an idea you want to pitch to someone. The magic wand version of this is you using telepathy

but we started with words. Imagining an experience is hard. Then low fidelity wireframes.

Figma made high fidelity prototypes easier, together with invision & co

But getting a feel for the experience was still up to the imagination

Prototypes were always better, but doing them required a designer to hook up wireframes and still design all the intermediary states

the ai to prototype is the closest we have so far to the magic wand solution.

[TB @tb8424]

ahh, finally time to relax
*opens phone to check the time*

[Alex @heyitsalexP]

Manager: "I want more visibility into our team's priorities"

Team: *builds out PM system, sends daily updates*

Manager: "Not like that! I can't spend all day reviewing tickets and roadmaps"

Team: *condenses everything into weekly update*

Manager: "Why are we doing project X?! I though I TOLD you our priorities are Y!"

Team: "It was in the weekly recap, I thought we agreed–if you had no feedback on the recap, it was ok to proceed with those projects"

Manager: "An email is going to get lost in my inbox and I don't have time to read a list of 50 projects anyway! Let's transition to a weekly 15 minute standup"

Team: ok, what time works for you?

Manager: "Mondays at 11a...except every second Monday of the month because that's my skip level. And every other week I have an executive team meeting that's usually at 10a but sometimes I get rescheduled to 10:30a and sometimes it runs over..."

If this sounds familiar, you need to confront two harsh truths:

1. You don't actually want "visibility", you want a team of mind readers, who will translate your fleeting thoughts into the perfect roadmap "make no mistakes".

This doesn't work for romantic relationships, and it doesn't work for managing teams. Grow TF up.

2. You're probably doing a piss poor job of setting priorities and communicating them.

If everything is an emergency, nothing is an emergency.

If the goal posts are always moving, building an action plan is impossible.

Stop managing your business reactively, from a place of fear, and take time to communicate your POV and priorities...and watch "management" magically get 10x easier.

Bonus: feeding your inbox and PM system into OpenClaw doesn't solve this problem because YOU are the problem.

[TB @tb8424]

Claude patting itself on the back

[Nan Yu @thenanyu]

If I had one big gripe with how we talk about these things, it’s that “design” is too narrowly conceived.

Experiment design is design. Chip design is design. API design is design.

We’ve boxed ourselves in so much that often I think designers don’t give themselves permission to think beyond visuals.

[John Loeber 🎢 @johnloeber]

you’re pitching garry tan

“so what do you guys do”

you start explaining

he’s furiously typing . two keyboards. one hand on each. you’ve never seen this before

“who are your top customers”

you explain. he types. his apple watch is a strobe light of notifications

“who’s your competition and why should i invest”

you explain that there’s no competition and you are the best and only product in the space

“false!” garry jumps out of his seat

“i am the competition!”

you are speechless

“in this meeting, i vibe coded your entire company. and my gstack has already closed your top customers.”

you check your phone. your stripe graph shows 100% churn

“and look at this”

garry shows you his imessage. there’s a text from 35 seconds ago. your top enterprise prospect that you’re trying to close? garry’s AI is trading baking recipes with the CEO’s mom

“thank you for playing!”

you have no moat. you are not admitted to the YC spring 26 batch.