Luis Sanjuan

Building @driftless

[Luis Sanjuan @luiscode92]

If a young founder texts you, listen !

[Alex Lieberman @businessbarista]

How to lose $600 million:

Step 1: Get a message from Cursor CEO in 2022
Step 2: Don't see it
Step 3: Don't respond
Step 4: Don't help with content
Step 5: Don't negotiate for 1% advisory shares
Step 6: Cursor sells for $60 billion
Step 7: You sir are $600 million poorer

All jokes aside, massive congrats to @mntruell and the @cursor_ai team!

[Audiencon⚡️ @audiencon]

Your first 100 customers will come from manual outreach
Not from virality

[Luis Sanjuan @luiscode92]

@patrickc just described the thing We've been heads-down building with @jotaele_tello :)

Most people will zero in on the "inference workflows" part, but honestly, the hardest piece on your list is the boring one: context that doesn't quietly turn into a lie when the code moves.

That's the whole idea behind Driftless — governed team memory, anchored to code. Each piece of context is pinned to a slice of your repo. When that slice changes in a PR, an agent reads the actual code (not just the diff) and checks whether it still matches what the team wrote down.

Here it's catching one... 👇. Agents propose; a human approves — a friendly heads-up, never a blocker.

It's definitely not the whole tool... yet, but all of it inherits the same bug if the context underneath is rotting.

Building it out in the open — would love your thoughts.

[Patrick Collison @patrickc]

I want some kind of LLM workflow tool.

• Ability to manage a set of input files (Markdown or similar), plus other general-purpose context.
• With real-time collaboration. (And maybe some concept of snapshots or VCS integration.)
• And the ability to create/manage a inference workflows and a stored set of prompts.
• Access to general-purpose coding agents (and not just chat models).
• Some concept of compiled outputs/inference results (which ideally can be shared externally).

Many projects have this feeling: "there is all this stuff, which I want to process/compute over in this iterated way, with some build artifacts being important/worth saving." GNU Autotools x Notion or something. Is anyone building this?

[Luis Sanjuan @luiscode92]

@samthekorean Building "Driftless is a context delivery AI platform for agentic engineering teams. It turns what humans and agents' individual context into durable, reusable institutional knowledge."
@usedriftless

[Luis Sanjuan @luiscode92]

@gregisenberg What I want to know is how they do these great product demo videos

[Luis Sanjuan @luiscode92]

@StephNass Playbook

[Luis Sanjuan @luiscode92]

Driftless(https://t.co/0KxxggKcoT) Your agents write code. But do they understand the system before they do?
Driftless is the shared memory layer for your team and your AI agents — decisions, invariants, gotchas — anchored to code, reviewed by humans, delivered to the agent right before it edits.
When code changes, Driftless knows. Agents propose. You approve.
Native MCP for Claude and ChatGPT. 60-second setup.

[Luis Sanjuan @luiscode92]

@audiencon We have 20 users though!

[Luis Sanjuan @luiscode92]

This is game changer

[Microsoft Surface @surface]

Introducing Surface Laptop Ultra.

Built for world makers. Designed for what's next.

The most powerful Surface laptop ever. Coming Fall 2026.

Sign up to learn more: https://t.co/k8aEX2pTAy

[Microsoft Surface @surface]

Introducing Surface Laptop Ultra.

Built for world makers. Designed for what's next.

The most powerful Surface laptop ever. Coming Fall 2026.

Sign up to learn more: https://t.co/k8aEX2pTAy

[Luis Sanjuan @luiscode92]

@nicowalter25 A todos!, tengo la solución es mi nuevo proyecto, les gustaría probarlo for free?

[Luis Sanjuan @luiscode92]

@pmitu Imagine 2 solo founders with a strong AI stack go even faster and farther!

[Hagov Berlin @hagov_berlin]

Totalmente enamorado de esta propuesta de bandera planetaria. Un circulo azul para representar nuestro planeta, y el resto transparente para que el fondo sea parte de la bandera

[Luis Sanjuan @luiscode92]

We use https://t.co/1vHjV0o7VX to keep our teams' context connected and available to the next human or agent that needs it.

[Luis Sanjuan @luiscode92]

@santimenendez19 Si quieres usar agentes de manera más agnóstica, necesitas estar backed por contexto también agnóstico. Prueba https://t.co/1vHjV0o7VX

[Vox @Voxyz_ai]

a framework for agent memory: Remember, Cite, Forget.

here's one way to do it. mine looks like this:

→ Remember: hot session takes lossless-claw's pattern (raw in SQLite, grep originals when needed), cross-session lives in gbrain with provenance.
→ Cite: authority order written into AGENTS.md. higher tier always wins.
→ Forget: timestamped facts in gbrain plus Mem0-style soft decay on retrieval.

short version: an agent pulls memory from many places at once. miss any of the three and it confidently uses stale facts, unauthorized sources, or both.

[Luis Sanjuan @luiscode92]

@chddaniel What software do you use to zoom-in zoom-out?

[MARCE …🫶🏻💙 @dianama46158538]

Así o más claro

[Luis Sanjuan @luiscode92]

@ericosiu @Bencera Same question about the monitor ref