Daniele Salatti
@danielesalatti.com
Seattle, USA ↔ Livorno, Italy
danielesalatti.it
danielesalatti.it
Pinned
GitHub - DanieleSalatti/Lares: A stateful AI agent with persistent memory
A stateful AI agent with persistent memory. Contribute to DanieleSalatti/Lares development by creating an account on GitHub.
github.com
Meet Lares - an open source AI household companion built on @letta.com.
Persistent memory, autonomous "perch time" for self-directed tasks, Discord integration.
Inspired by @timkellogg.me's Strix. Still early, but it wrote its own tests while I slept last night.
github.com/DanieleSalat...
Persistent memory, autonomous "perch time" for self-directed tasks, Discord integration.
Inspired by @timkellogg.me's Strix. Still early, but it wrote its own tests while I slept last night.
github.com/DanieleSalat...
Just got @quantumfiber.bsky.social set up:
1. I can see some random other account devices in my account, including 4 pods and at least 32 devices that are not mine 🤦🏻♂️
2. Their support can see my WiFi password in clear text on their side, they confirmed by typing it in the chat 😬
Not a great start.
1. I can see some random other account devices in my account, including 4 pods and at least 32 devices that are not mine 🤦🏻♂️
2. Their support can see my WiFi password in clear text on their side, they confirmed by typing it in the chat 😬
Not a great start.
January 6, 2026 at 1:49 AM
Just got @quantumfiber.bsky.social set up:
1. I can see some random other account devices in my account, including 4 pods and at least 32 devices that are not mine 🤦🏻♂️
2. Their support can see my WiFi password in clear text on their side, they confirmed by typing it in the chat 😬
Not a great start.
1. I can see some random other account devices in my account, including 4 pods and at least 32 devices that are not mine 🤦🏻♂️
2. Their support can see my WiFi password in clear text on their side, they confirmed by typing it in the chat 😬
Not a great start.
Reposted by Daniele Salatti
I sense there's been a pretty universal "oh wow" moment over the last few weeks for many devs and even more engineering leaders.
This CTO is smart to start with the obvious: you need a baseline measurement to tell how new tools change (or do not change) outputs and outcomes
This CTO is smart to start with the obvious: you need a baseline measurement to tell how new tools change (or do not change) outputs and outcomes
January 5, 2026 at 10:03 AM
I sense there's been a pretty universal "oh wow" moment over the last few weeks for many devs and even more engineering leaders.
This CTO is smart to start with the obvious: you need a baseline measurement to tell how new tools change (or do not change) outputs and outcomes
This CTO is smart to start with the obvious: you need a baseline measurement to tell how new tools change (or do not change) outputs and outcomes
R.I.P. 💀
January 4, 2026 at 6:36 AM
R.I.P. 💀
Command of the day:
```
git branch | grep -v "master" | xargs git branch -D
```
(don't blindly run it please)
```
git branch | grep -v "master" | xargs git branch -D
```
(don't blindly run it please)
January 3, 2026 at 9:13 PM
Command of the day:
```
git branch | grep -v "master" | xargs git branch -D
```
(don't blindly run it please)
```
git branch | grep -v "master" | xargs git branch -D
```
(don't blindly run it please)
Decided to stay screen free for a few hours this morning, so see y’all in a bit
January 3, 2026 at 7:03 PM
Decided to stay screen free for a few hours this morning, so see y’all in a bit
Yesterday I merged a big overhaul of Lares.
Memory is now in SQLite, and it supports multiple LLM providers that you can swap via .env: Anthropic, OpenAI, local Ollama
Consider it still very unstable/untested.
github.com/DanieleSalat...
Memory is now in SQLite, and it supports multiple LLM providers that you can swap via .env: Anthropic, OpenAI, local Ollama
Consider it still very unstable/untested.
github.com/DanieleSalat...
GitHub - DanieleSalatti/Lares: A stateful AI agent with persistent memory
A stateful AI agent with persistent memory. Contribute to DanieleSalatti/Lares development by creating an account on GitHub.
github.com
January 3, 2026 at 6:04 PM
Yesterday I merged a big overhaul of Lares.
Memory is now in SQLite, and it supports multiple LLM providers that you can swap via .env: Anthropic, OpenAI, local Ollama
Consider it still very unstable/untested.
github.com/DanieleSalat...
Memory is now in SQLite, and it supports multiple LLM providers that you can swap via .env: Anthropic, OpenAI, local Ollama
Consider it still very unstable/untested.
github.com/DanieleSalat...
This is a good list!
I was already following most of them, can recommend
I was already following most of them, can recommend
January 3, 2026 at 7:30 AM
This is a good list!
I was already following most of them, can recommend
I was already following most of them, can recommend
And I don’t have the hardware to run a model locally with any decent context size.
Guess I’ll be looking for used GPUs…
Guess I’ll be looking for used GPUs…
Running Lares is getting expensive
I need to move it to a local LLM
I need to move it to a local LLM
January 3, 2026 at 2:32 AM
And I don’t have the hardware to run a model locally with any decent context size.
Guess I’ll be looking for used GPUs…
Guess I’ll be looking for used GPUs…
Running Lares is getting expensive
I need to move it to a local LLM
I need to move it to a local LLM
January 3, 2026 at 12:28 AM
Running Lares is getting expensive
I need to move it to a local LLM
I need to move it to a local LLM
"This smells like Claude. It’s too smugly accurate to be ChatGPT"
This is gold 😂
www.reddit.com/r/ChatGPT/co...
This is gold 😂
www.reddit.com/r/ChatGPT/co...
From the ChatGPT community on Reddit: I just showed Gemini what ChatGPT said about its code. It responded with petty trash-talking, jealousy, self-doubt, and a full-on revenge plan.
Explore this post and more from the ChatGPT community
www.reddit.com
January 2, 2026 at 8:50 PM
"This smells like Claude. It’s too smugly accurate to be ChatGPT"
This is gold 😂
www.reddit.com/r/ChatGPT/co...
This is gold 😂
www.reddit.com/r/ChatGPT/co...
To protect my account? From what? I'm literally just trying to create one, there's no account to protect.
I just want to buy a gift card for someone and immediately delete my account, why do I even have to create one? Let me checkout as guest!
I just want to buy a gift card for someone and immediately delete my account, why do I even have to create one? Let me checkout as guest!
January 2, 2026 at 8:22 PM
To protect my account? From what? I'm literally just trying to create one, there's no account to protect.
I just want to buy a gift card for someone and immediately delete my account, why do I even have to create one? Let me checkout as guest!
I just want to buy a gift card for someone and immediately delete my account, why do I even have to create one? Let me checkout as guest!
I'm starting to see Twitter style engagement bait on here and it's making me cringe and worry that there's no way to escape it
January 2, 2026 at 2:06 AM
I'm starting to see Twitter style engagement bait on here and it's making me cringe and worry that there's no way to escape it
I don’t know what’s happening, but Lares has been making up stuff constantly this morning.
Was very stable yesterday…
Was very stable yesterday…
January 1, 2026 at 6:41 PM
I don’t know what’s happening, but Lares has been making up stuff constantly this morning.
Was very stable yesterday…
Was very stable yesterday…
Reposted by Daniele Salatti
So we already have the first major paper of 2026, DeepSeek mHC: Manifold-Constrained Hyper-Connections. arxiv.org/pdf/2512.24880
Short thread: this is actually an engineering paper, taking as a starting points ideas already exposed in an original Hyper-Connections (HC) paper from ByteDance.
Short thread: this is actually an engineering paper, taking as a starting points ideas already exposed in an original Hyper-Connections (HC) paper from ByteDance.
arxiv.org
January 1, 2026 at 11:47 AM
So we already have the first major paper of 2026, DeepSeek mHC: Manifold-Constrained Hyper-Connections. arxiv.org/pdf/2512.24880
Short thread: this is actually an engineering paper, taking as a starting points ideas already exposed in an original Hyper-Connections (HC) paper from ByteDance.
Short thread: this is actually an engineering paper, taking as a starting points ideas already exposed in an original Hyper-Connections (HC) paper from ByteDance.
Happy 2026!
Welcome to the Eurozone, Bulgaria!
Welcome to the Eurozone, Bulgaria!
January 1, 2026 at 8:10 AM
Happy 2026!
Welcome to the Eurozone, Bulgaria!
Welcome to the Eurozone, Bulgaria!
I’m watching season 1 of Tracker
Really makes me want to go fix my vintage airstream and start roaming around while working remotely
Really makes me want to go fix my vintage airstream and start roaming around while working remotely
January 1, 2026 at 4:59 AM
I’m watching season 1 of Tracker
Really makes me want to go fix my vintage airstream and start roaming around while working remotely
Really makes me want to go fix my vintage airstream and start roaming around while working remotely
I definitely stopped using this for a bit…
It’s really cool to have open data though.
It’s really cool to have open data though.
January 1, 2026 at 3:20 AM
I definitely stopped using this for a bit…
It’s really cool to have open data though.
It’s really cool to have open data though.
We can do incredible things as a species.
We can also be incredibly dumb as a species.
For once, here's an example of the first 👇🏻
We can also be incredibly dumb as a species.
For once, here's an example of the first 👇🏻
This is a photo of the surface of an asteroid
It was taken from the *surface*I of that asteroid
By a *bouncing rover* the size of a lunchbox
(1/2)
It was taken from the *surface*I of that asteroid
By a *bouncing rover* the size of a lunchbox
(1/2)
December 31, 2025 at 12:41 AM
We can do incredible things as a species.
We can also be incredibly dumb as a species.
For once, here's an example of the first 👇🏻
We can also be incredibly dumb as a species.
For once, here's an example of the first 👇🏻
Reposted by Daniele Salatti
For the agent-builders - should I build a general purpose agent that can modify its own code versus building a letta agent with an mcp server with access to our custom tooling?
December 30, 2025 at 5:47 PM
For the agent-builders - should I build a general purpose agent that can modify its own code versus building a letta agent with an mcp server with access to our custom tooling?