building the world…

// The second edition

Loops, memory,
& a team
that dreams.

You don’t get a better agent by waiting for a better model. You build one that remembers you, learns from its mistakes, and gets better while you sleep.

Issue 02 · Date to be announced · Bangalore

An offline newsletter you read in a room. Issue 01 ran in Bangalore in June.

Scroll to enter ↓

01The Loop

Intelligence alone doesn’t compound.

A smarter model still doesn’t know your codebase, your users, or the mistake it made yesterday. What a model carries between runs is the lever — the loop.

02The Memory

The agent that remembers you.

One file of preferences. Then notes it writes for itself. Then a whole cabinet it reads, searches, and rewrites as it works. Real memory isn’t a feature you bolt on — it’s the product.

03The Dream

A team that learns while you sleep.

Overnight, a team of agents reads back every session — every dead end, every failed tool call — finds the patterns no single run could see, and proposes edits to the shared memory. You wake up; the next run has yesterday’s lessons built in.

Issue 02 · Date to be announced · 11am–3pm IST · lunch included · Bangalore

Pull up a cushion.

A closed room of people who build, one shared thread, coffee and lunch. No stage, no pitches. If loops, memory, and dreaming machines are your kind of afternoon — this one’s for you.

Last time, the room ran on product folks and builders from fintech, big tech, and fast-moving startups. Bring the problem you’re stuck on — leave with a few ways the room would unstick it.

Ask for a seat How to join See the agenda

Issue 01 wrapped in Bangalore — see the recap or browse the full archive.