Nik

Building Nodo - Text it, forget it. Nodo follows up.

[Nik @nikhil17]

Product management is being rebuilt around AI.

The winners will not be PMs who use ChatGPT for random tasks.

The winners will be PMs and teams that turn AI into repeatable workflows for discovery, prioritization, PRDs, prototyping, handoff, testing, and learning.

[Nik @nikhil17]

Finally implemented read receipts and typing indicators (3 dots "...") in Nodo.

Crazy how much more human it makes the experience.

Sometimes better AI isn’t just the model.

It’s the tiny UX details that make it feel alive!

[Nik @nikhil17]

With a good harness an AI developer agent is better than an average engineer

[Nik @nikhil17]

𝗧𝗵𝗲 𝗯𝗶𝗴𝗴𝗲𝗿 𝗹𝗲𝘀𝘀𝗼𝗻 – Focus on the culture and the harness

The most striking part of this playbook - only one of these items is about technology (#2). The rest are org design, culture, hiring, and incentive design.

Most enterprises are spending all their energy on the platform and almost none on the operating system around it.

[Nik @nikhil17]

@tryramp just raised $750M at a $44B valuation.

But the story isn't the fundraise. It's what's underneath it:
- Two-thirds of Ramp's code is now written by AI
- 99.5% of employees are active on AI tools
- 84% use coding agents weekly
- Non-engineers account for 12% of all human-initiated pull requests on the production codebase — thousands per month.

The result – 70+ products shipped in a single quarter.
Here's how they did it (credit: CPO @geoffintech )

[Nik @nikhil17]

𝟰/ 𝗧𝗵𝗲𝘆 𝗿𝗲𝗺𝗼𝘃𝗲𝗱 𝗲𝘃𝗲𝗿𝘆 𝗰𝗼𝗻𝘀𝘁𝗿𝗮𝗶𝗻𝘁

No token limits. No tiered access by role. No connector requests sitting in a queue. Unlimited budget to build, learn, and explore.

Geoff Charles framed the cost math simply: "If someone is 2x more productive with AI, you should be willing to spend their entire salary again in tokens."

[Nik @nikhil17]

@Railway This is insane. Surprised you don’t have a backup given the scale you’re operating at

[Nik @nikhil17]

As a University of Michigan alum, this one made me smile.

Michigan’s endowment reportedly invested $20M in OpenAI very early.

That position is now worth~$2B.

A 100x return.

Most institutions access early-stage startups through venture funds.

Michigan appears to have gone direct - taking concentrated risk on a company that was far from the obvious winner it is today.

That takes conviction.

The best asymmetric bets usually look a little unreasonable before they look obvious.

Go Blue!

[Nik @nikhil17]

🚀 My weekend project is now open in beta.

Meet Nodo — an AI accountability buddy inside WhatsApp.

Text it tasks, ideas, reminders, goals, links, or random thoughts.

Nodo organizes them, helps prioritize, and follows up so things don’t slip.

I built it because I got tired of losing important thoughts in notes apps, screenshots, group chats, and my own head.

Most productivity tools wait for you to come back.

Nodo follows up.

I’ve been using it myself for 2 months and it’s finally ready for a small group of beta users.

Would love your feedback:

https://t.co/DELLTqaHVl

[Nik @nikhil17]

When Claude yells at you for going down a rabbit hole

[Nik @nikhil17]

$GOOGL on a tear 🚀

Stock up 9% today, 1.5x in 1 year

Startup-like growth numbers as a publicly traded behemoth… crazy

[Nik @nikhil17]

5/ The bigger question:

What other domains look like software engineering?

Where is the work valuable, repeatable, high-context, and objectively measurable?

That may be where the next AI behemoth gets built.

[Nik @nikhil17]

Anthropic’s original pitch was simpler than people think:

Turn inference into a compounding loop.

Lets dive in 👇

Credit: @AnjneyMidha’s CS 153 Class: Frontier Systems

[Nik @nikhil17]

4/ The model acts.

The environment responds.

The system learns.

That’s the loop.