It feels like the builder community has adopted a new rule: build in public.

At this point it almost feels like a social norm and anyone who chooses not to is often looked down on.

In fact, @brian_armstrong recently tweeted:

Operating in stealth mode is almost always a mistake.

Talk publicly about what you're building. You'll build momentum, get real feedback, and someone will reach out with the other half of your idea you didn't realize you were missing.

— Brian Armstrong

When I saw this tweet, it made me reflect on why I chose stealth in the first place to build Backtick (not working on it anymore). It was in stealth mode for its entire lifespan. And honestly, one reason convinced me to do that.

We knew the problem, but not the solution

Backtick started with a simple observation: Most issue trackers are just glorified to-do apps. (This is hard for Linear fanboys to digest, but it's true.)

If coding IDEs like Cursor can start writing code for us, why can't issue trackers automatically track and manage work?

Great pitch.

But what does that actually look like in reality?

We had no idea. So we started searching for the answer.

Iteration #1: More AI automation

Nope.

Too much automation kills productivity more than no automation.

Everyone wants the "sweet spot" of automation, but no one actually knows where that sweet spot is — not even the teams asking for it.

Iteration #2: Cursor for project management

So… a chat window where you ask questions about your tasks? Sounds cool.

But what does that actually mean in practice? No one really knows.

YC even had an RFS called "Cursor for Product Managers."

Two thoughts on that:

1. Product managers usually don't buy software. They rarely have purchasing power.
2. Whenever investors say "we're looking to fund this exact idea" — run.

Iteration #3: AI agent that behaves like a junior PM

This is where we had the most success.

We built logic that tried to understand:

• the product
• its stage
• market dynamics
• competitors

Then the system would:

• prioritize tasks
• generate tasks from customer feedback
• assign work to the right people
• keep status updated

Think of it as:

If Productboard and Linear had a baby — but with superpowers.

And honestly? It worked surprisingly well for simple B2B SaaS products.

But it broke when we tried to scale the idea.

Product strategy isn't universal.

• Social consumer products behave differently from fintech consumer products.
• Two companies in the same industry can have completely different strategies.
• Even direct competitors can have completely different product plans.

After 8 months we learned something simple:

Most product strategy comes from intuition.

To give that intuition to our AI agent, we would basically need to build consciousness. And we weren't ready to build AGI with $200k in pre-seed funding.

So we went back to zero. Every version of our solution had limitations.

We weren't sure if the solution made sense now — or ever.

The point

Stealth mode isn't BS. It actually makes sense when you're working on a new idea and aren't sure what the solution looks like yet.

Some great companies stayed in stealth for years: Stripe, SpaceX, Palantir.

The main argument for building in public is early feedback. But early customer feedback can also be misleading when the product category itself isn't well understood.

One of my jobs as a founder is to minimize regret. The way I try to do that is by failing on my own terms.So if you want to build in stealth — do it. If you want to build in public — do it.

Just don't do it because it's trendy or because some successful founder tweeted about it.