Forem: elend80

Google Gemini 3 Pro — Key Features, Ecosystem Updates, and Technical Evaluations

elend80 — Fri, 21 Nov 2025 18:37:53 +0000

Google’s Gemini 3 Pro represents a clear shift in Google’s AI direction — moving toward deeper reasoning capabilities, long-context understanding, and more autonomous agentic workflows across the ecosystem.

I curated a collection that brings together the most useful resources about Gemini 3 Pro:

Core model features
Ecosystem releases
Technical evaluations
Tooling updates
Early benchmarks and community analysis

📚 Full curated collection here:

👉 https://www.culink.io/teamculink/gemini-3-pro

Whether you're exploring reasoning models, building agents, or researching multimodal systems, this collection gathers the most helpful links in one place.

If you’ve come across other high-quality Gemini 3 resources, feel free to recommend — I’ll keep expanding the list.

Building 100 Link Collections in 30 Days — Day 7: What I Know (and Don't Know)

elend80 — Thu, 06 Nov 2025 14:53:17 +0000

The Challenge

I'm building Culink, a platform for curated
link collections (think Pinterest for useful
resources instead of images).

To find product-market fit: create 100
collections in 30 days.

Day 7 Numbers

What I know for certain:

22/100 collections created
200 visitors (Show HN + IH)
14 signups (7% conversion)
4 created collections (29% activation)

Topics: AI/ML, startups, dev tools,
productivity, economics, lifestyle.

Decent metrics for 7 days.

But here's the problem.

What I Don't Know

These numbers tell me almost nothing about
product-market fit:

I don't know:

Which topics actually resonate
Why people signed up
Why most didn't create collections
What would make them come back
Who my target audience should be

200 visitors is too small to see real patterns.

14 signups is too few to validate anything.

4 creators? That's just noise.

The Temptation to Overinterpret

With small numbers, it's easy to see patterns
that aren't there.

"AI collections got more clicks!"
→ Or did they just appear first?

"People prefer browsing to creating!"
→ Or is the creation flow too complex?

"Tech people are my audience!"
→ Or is that just Show HN selection bias?

I could make up stories to explain these
numbers.

But that would be guessing, not learning.

What I'm Using ChatGPT For

ChatGPT helps me create collections faster:

Suggests relevant resources
Writes descriptions
Organizes by category

It's good for speed and structure.

But it can't tell me:

Who needs this
What problems they have
Whether this solves them

That requires talking to real people.

What Culink Does

Create collections by topic.
Share publicly or privately.
Follow other curators.

Early stage. Rough edges.

→ culink.io/discover

What I Need to Learn

Before making 78 more collections,
I need to answer:

Who curates content regularly?

Newsletter writers?
Developers (awesome lists)?
Researchers?
Someone else?

Why do they curate?

For their audience?
Personal organization?
Building authority?

What's their biggest problem?

Finding quality sources?
Verification?
Organization?
Keeping it updated?

The Plan

Week 2:

Interview 5 people who actually curate
Understand their real workflows
Find common pain points
Pick one target segment
Build for them specifically

No more guessing.
No more interpreting small numbers.

Just direct conversations with real curators.

If You Curate Content

I'm looking for 5 people to interview
(15-20 min each).

If you regularly curate — newsletter,
awesome lists, research collections,
anything — I'd love to learn from you.

What you curate, your workflow, your
challenges.

Comment below or reach out.

Building in Public

This is Day 7 of 30.

The honest version: I have metrics but
no clarity.

The numbers look okay, but I still don't
know who I'm building for.

That's what Week 2 is for.

Day 7/30
www.culink.io/discover
Twitter: @byun_yh

I built Culink — a simple tool to organize and share meaningful links

elend80 — Wed, 29 Oct 2025 15:41:43 +0000

Hey everyone

I’ve been working on Culink for the past few months as a solo developer from Korea.

It’s a minimal web app to organize and share meaningful links — kind of like Pinterest for links (articles, tools, videos, and resources).

Why I built it
I kept losing interesting links across messy bookmarks.

GitHub “awesome lists” are great but static — I wanted something more visual, social, and collaborative.

What it does

Create themed collections (AI tools, dev resources, startup guides, etc.)
Collaborate and share collections with others
Follow curators you like
Discover quality content through your network

Tech stack

Next.js 15
NestJS
PostgreSQL
Azure Container Apps

Built solo over a few months 🇰🇷

Try it
You can explore it here: https://www.culink.io/discover

I’d love your thoughts

What use cases come to mind for you?
Anything confusing in the UX?
Features you’d want before adopting it?

Thanks for reading!