Parul Singh

A pretty bold comment in Nature written by linguists, computer scientists and philosophers declaring that AGI has been achieved.

"By reasonable standards, including Turing’s own, we have artificial systems that are generally intelligent. The long-standing problem of creating AGI has been solved."

After executing Alex Pretti on Saturday, ICE has shifted to primarily targeting bus stops and elementary schools in my community this week. I cannot overstate how much of the ICE activity in my community is just targeting places where little kids are.

A bird's-eye view of a former Auschwitz II-Birkenau camp showing a wide dirt pathway flanked by parallel rows of barbed-wire fences. Groups of visitors walk along the path, surrounded by the remnants of brick structures and barracks, now reduced to foundations. Green grass contrasts with the somber history of the site, as the path leads toward a guard tower in the distance.

Auschwitz was at the end of a process. We must remember that it did not start from gas chambers.

This hatred gradually developed: from ideas, words, stereotypes & prejudice through legal exclusion, dehumanization & escalating violence... to systematic and industrial murder.

Auschwitz took time.

6. That’s why the stickiest AI products aren’t “AI-first.”
They’re *problem-first*.

📨 I unpacked this idea in my latest newsletter: parul.substack.com/p/falling-in...
It’s about how falling in love w the problem helps AI products win.

What’s one AI tool that erased real pain for you?

5. That’s when AI clicks.
Not when you show what the model can do.
But when the user says:
“Oh. I never want to do this manually again.”

4. Builders who get this ask better questions:
✦ Where are users duct-taping a workflow today?
✦ What do they copy/paste 10 times a day?
✦ What feels like magic *only because* it was previously awful?

3. That wedge only shows up when you fall in love with the problem.

Not the model.
Not the market.
The actual problem—at the atomic level of real users, real pain, real friction.

2. These products are sticky not because of their tech.
But because they nailed the wedge.

They found one painful, valuable, *frequent* task.
And made it vanish.

1. The best AI tools don’t dazzle.
They disappear.

They compress 15 clicks into 1.
Auto-fill your messy CRM.
Draft the response you *meant* to write.
Catch the error you missed.

🧵 The stickiest AI products don’t feel magical.
They feel inevitable.

They don’t just show off what the model can do.
They erase something frustrating from your workflow—so smoothly, you can’t imagine going back.👇

7/
Prompt chains aren’t just for productivity.

They’re a new way of thinking—at founder speed, with strategist depth.

💬 If you're building this way, what chains have you tried? What are your most useful strategy prompt hacks?

6/
And as tools evolve, this gets even more powerful:

→ Agents that remember your context
→ Embedded copilots in your dashboards
→ AutoGPTs that can run the whole chain for you

But the mindset shift matters:
🔗 Think in chains.
🍰 Refine in layers.

5/
🧭 Here’s a 5-prompt strategy chain I run monthly w founders :

1. What changed this month?
2. Are we differentiated enough from [X competitor]?
3. What would a skeptical buyer distrust on our site?
4. What’s fragile in the business rn?
5. What user behavior could unlock LTV?

4/
That’s the power of a *prompt chain*.

It’s not about output.
It’s about structured, layered thinking.

When done right, a prompt chain becomes something like a *mindflow* — a way to deepen clarity, test hypotheses & find sharp answers *faster* than any brainstorming session.

3/
It became an instant mini-diligence loop.

Each follow-up prompt refined the strategy.
The ideas got better.
The positioning got sharper.
I walked away with a spreadsheet I could hand to my product team—or my favorite vibe coder.

All in a single conversation.

2/
I asked ChatGPT for 10 startup ideas:
✔️ Based on verticals I care about
✔️ With product-led growth potential
✔️ Subscription pricing $250–$500/mo

Then I kept going...

What keywords would matter?
Could we win via SEO?
What’s the fastest GTM motion?

1/
🔗 Founders are sleeping on one of the most powerful uses of AI right now:
Prompt chains. 🔗

Not for writing emails or fixing grammar.
But for thinking. Strategically.

Here’s how I used one to generate startup ideas, GTM plans, SEO strategy—and more—in under an hour 🧵

Founder myth tends to reward charisma and momentum. But behind the scenes, real mastery looks like clarity, compounding, and staying calm in the hard moments.

Who’s a founder you admire, and what have they taught you? (3/3)

They’re the ones who’ve quietly shipped through chaos, built exceptional cultures, and made it look effortless when it wasn’t. (2/3)

Name a founder more people should be learning from—someone whose name doesn’t get mentioned enough.

I’ve been lucky to back and learn from some remarkable founders. The ones I keep thinking about aren’t always the loudest or most followed. (1/3)

(3/3) It’s not just “busywork.” It’s the glue -- the repetitive, context-switch-heavy work that slows down execution across the org.

What are you handing off first? Bonus: for your favorite tools (or are you creating bespoke ones?)

(2/3)
The top answers?

💻 Coding
📆 Scheduling & email triage
🧑‍💻 Document processing & data entry
📝 Content creation & editing
💪🏽 Automating workflows in onboarding / customer success
☑️ Task management
🧐 Research & decision support

If AI could take one thing off your plate today -- what would you hand over without hesitation?

I’ve been asking this in 1:1s with founders. And what’s surprising isn’t that they’re automating more -- it’s what they’re automating first. (1/3)

congrats!!! 🎉

If you're building lean, embedded, and impossible to rip out — you’re not alone.

This is what survival looks like in 2025.

Thanks to @henrythe9ths for compiling the Lean AI Leaderboard.

(10) They execute like survival depends on it — because it does.
No tourists. No hype-chasers. Just sharp operators with early revenue and fast iteration loops.

(9) Infra isn’t the pitch.
It’s just what lets the product scale.

(8) Vertical AI is outperforming.
Legal, healthcare, logistics. Regulated = defensible.

(7) Optionality > overcapitalization.
Many never raised a priced round. Some are bootstrapped.