I couldn’t find something like this here. If y’all have an existing platform, please share.

What this is:
- A safe discussion space for PMs to discuss their work, growth, and rant about stakeholders being oblique, managers being toxic, etc.
- Discuss new things in tech, AI, startups etc.
- Ask work related questions they’d be uncomfortable asking their colleagues or friends.

What this is not:
- A community where people request for referrals or ask for openings; asking for advice is okay.

In times of work-related insecurities and instabilities, I feel this would be a good idea for everyone to come together and discuss what’s going to impact us the most. Invite-only platform and work credentials are mandatory for receiving an invite. Will make a discord or WA chat.

Edit: DM me for the invite!

I’m thinking of starting a habit of reading 1 research paper a week on Product Management.

In engineering, there seems to be a very established culture around reading papers, following research, discussing findings, etc. But in Product Management, I rarely come across PMs talking about research papers as part of their regular learning.

I’m not looking for the usual PM books, LinkedIn posts, frameworks, or case studies.

I’m interested in research that helps us step away from the day-to-day execution and think more deeply about things like:

Product strategy & decision-making

Customer behaviour

Product discovery

Innovation & new product development

Prioritisation

Product-market fit

Team/product organisation

Product experimentation

AI & product management

I recently discovered that there is actually a large body of research under New Product Development, Product Innovation, and Innovation Management. For example, there are decades of academic research on product development and innovation.

So I’m curious:

1. Do you already follow/read research papers as a PM?

2. Where do you find good papers?

3. Are there specific journals, researchers, newsletters, communities, or websites you recommend?

4. Would anyone be interested in doing something like “1 paper/week” and discussing the practical implications for PMs?

I’m particularly interested in research that can make us better product thinkers, not just better at executing the PM process.

finally, i was able to replicate this kind of 3d interactive world in the browser itself. it runs completely well even on mobile.

i don't have any game dev experience and of course I'm not a graphics engineer, but the level of capabilities these models have opened up is just mind boggling.

reddit only allows you a GIF of less than 100 MB, so I had to clip out most of the video:(

tech stack i used for this
---------------------------------------------------------

• three.js for the 3d scene
• react + vite for the frontend
• opus 5.5 (high) for the code
• supabase for the database
• vercel for deployment

the process i used. you can build for your website
----------------------------------------------------------

step 1: find a reference for the website layout

• do not start from a blank page. better if you have a website ref to start with.
• if not ask claude to give you 3 radically dif websites using mcp tools like mobbin
• pick the one closest to your idea. i use a design iterative process i found from lenny.
• btw i sketched my hero by hand and asked claude to turn it into a wireframe

step 2: set the design and art direction

• lock the look before you write any code. try /design in claude to iterate with mocks
• use reference images to generate art style or direction with chatgpt or google image models
• do the back and forth in the image model. it is cheap to iterate there, and expensive in code

step 3: build the main objects (ship, background, people)

• opus is very good at real world 3d objects but very not good at modelling humans yet.
• it does not really sculpt a person. tried it multiple times but gave me slop.
• so for humans: i did not build them in code. i downloaded a free 3d modelled pack. built the clothing and hair on top of it in code.
• for the ship and background: i generated sprite sheets for each in the image model. this keeps claude close to your vision, with no drifting and no long back and forth

step 4: build the 3d scene phase by phase

• my 3d reference was meng to's "the river voyage" site.
• i built one scene per phase (i have 4 phases or scenes in my website: profile, applications, mocks, pod prep). don't do it all at once if you have multiple scenes. it'll hallucinate.
• the start takes a lot of steering. once the base is right, updating the graphics is much easier

step 5: let claude rate its own work
(this is the best trick i found and it saves time & effort)

• ask claude to spawn multiple agents that score the visuals in loop until a benchmark
• for eg prompt: "perfect the ship shape, textures and physics until it matches the sprite sheet and gets an overall rating of at least 8 out of 10. spawn multiple agent to do so"
• do it one element at a time: ship, people, background, water, and so on
• more time & rounds give better graphics. you decide how far to push.

step 6: design the page ui before you code it
(this is a continuation of step 1 or you can fork it for the 3d scene & website seperately)

• use /design and ask for mockups first
• ask for 3 radically different options, then pick the one that fits
• i followed the design workflow from lenny's article: "How to turn your AI into a world class designer "

step 7: stitch it all together

• ask claude to merge the graphics layer and the ui into one site
• run it locally and review it first. test it for all screen types (mobile & laptop)
• iterate until it feels right. main part is also optimizing performance (don't skip)

step 8: merge & deploy

• ask claude to push the code to github first. helps you maintain the version history
• then deploy. i used vercel for hosting and supabase for the database.
• If you have a domain purchased, set it up and ask claude in chrome to configure all DNS settings.

it took me around 5-6 hours with multiple sittings to completely finish it. can be done faster as well if you know the flow. still would recommend spending time to perfect it.

Rish from productminds :)