Christian M.M. retweeted
Same task, two assistants:
Spettro Chat vs Gemini.
Explain the pi algorithms, show how fast they run, how accurately, and generate interactive artifacts that show it.
Which one did a better job?
Imo, Spettro did
Christian M.M. retweeted
All this talk about models "escaping the sandbox" rn is straight up horsesh*t.
These labs aren't dealing with rogue superintelligence.
They just have misconfigured testing environments and want the PR hype.
If you want to make a hardened sandbox, you don't rely on model guardrails.
On Spettro (spettro.app), we built real kernel-level sandboxing for the harness.
Here is how you actually lock an LLM down:
- default deny: all commands trying to reach outside the box are automatically blocked.
- network limits: hard constraints on exactly what domains, ports, and requests are allowed.
- folder isolation: the model is locked to specific work folders.
It physically can't touch anything else.
An LLM literally cannot "vibe" its way out of an OS-level constraint.
Unless there's an unpatched zero-day in the kernel, exploiting it is 100% impossible.
Stop falling for the marketing hype.
Build better sandboxes.
Christian M.M. retweeted
100% of the launchpad fees dedicated to buying back and burning $SOLCEX. Imagine that. 🔥
Christian M.M. retweeted
I can finally say that I will be there, bringing spettro with me
See you in Ireland!
Bears are being chased out of Ireland.
@solana builders, join CastleDAO at the historic Slane Castle
Two weeks of deep focus, ambitious ideas and the right people, in a literal castle.
24 August • September 6
Apply now ↓
Christian M.M. retweeted
This chain is impossible sometimes.
What is actually bullish to people?
@DaemonTerminal has been tweeted by Toly.
It’s been pushed by a community of thousands.
@ricomaps is, in my opinion, the best value Solana forensics platform available.
Yet all I seem to hear is that I’m not doing enough.
So tell me.
Don’t just ask “what’s the plan?”, “what’s next?”, or “where’s the dev?”
Give me actual feedback.
Use the products.
Tell me what feels broken.
Pitch ideas.
Offer to help.
Tell me what would make you use them every day.
There hasn’t been a tech project that’s really broken out on Solana in months.
Every time I sit down and think about what to build next, I end up at the same conclusion…
Daemon is the thing.
I genuinely believe Daemon should be massive.
But the reality is there aren’t many users, I don’t get much feedback, issues rarely get reported, and very few people tell me what they actually want.
So what do you want me to build?
Genuinely.
I’m not looking for “soon” culture or empty hype.
I’m looking for real opinions from the people who actually rock with me.
What would make Daemon or RicoMaps something you couldn’t live without?
I’m listening.
We’ve been quiet lately, and we apologize but it’s been for a good reason.
We’ve been locked in building RicoMaps V2.
As the market begins to turn, we used this slower period to rebuild, refine, and make sure the product is ready for what’s ahead.
RicoMaps V2 launches this week.
We’re keeping most of what’s coming a surprise, but trust us this is going to be massive.
Christian M.M. retweeted
Almost 6 years of building apps and I'm still torn on the best way to offer trials.
Users & Devs:
What is your preferred model when trying out a new desktop or mobile app?
Do cheap paid trials work, or is free the only way to go?
67%Free Trial 7-14 days
19%Cheap paid Trial
11%No trial
4%None (comment)
27 votes • Final results Christian M.M. retweeted
Finally, I can code from my bed directly through Spettro for iPhone and iPad
I love this feature
Christian M.M. retweeted
Huge shoutout to the team at @alibaba_cloud!
We just secured a massive discount on GLM 5.2, SuperSmart and Kimi K2.7 Code.
Lower inference costs mean your Spettro subscription now lasts even longer, giving you far more execution power and longer coding sessions for the exact same price.
Enjoy the extra runway.
Still building.
Christian M.M. retweeted
Brainblast has been downloaded over 8,500 times by developers around the globe.
It’s a powerful tool that helps detect errors that often get shipped unnoticed by most AI.
But it can still be so much more.
Soon, it will be. 🧠💥
Christian M.M. retweeted
I signed the Open Weights & American AI Leadership letter — signature #81 as Carlo Esposito.
ow-b16186
openweights.gitlawb.com/s/81
Thanks @1kartikkabadi1 for the suggestion
Christian M.M. retweeted
Back into iOS & macOS development, and it’s hitting me with major nostalgia from my Apple Developer Academy days.
Hot take: I still prefer Kotlin over Swift for mobile dev.
But Swift + SwiftUI are incredible, moving away from the old UIKit setup makes building for macOS actually enjoyable now.
Plus, building with AI in 2026 feels like a superpower compared to how painful it used to be.
To celebrate, I’m building Spettro UI for macOS using the CLI version of Spettro itself.
Is this AGI? (jk)
Christian M.M. retweeted
Guys, the bear market is in full effect, and Robinhood is pulling even more liquidity and attention away from Solana.
Those things are out of my control. Trust me, it hurts me just as much as it hurts you to watch the market move like this.
All I can do is keep building and that’s exactly what I’m doing. Updates are still shipping, work is still happening, and a lot is moving behind the scenes.
Markets move in cycles. Look at Solana’s highs and lows over the years. Recovery may not happen as quickly as any of us want, but when the bulls return, the projects that kept building will be ready.
$DAEMON. $RICO. $LCKD.
We’re not going anywhere.
Christian M.M. retweeted
AI coding tools tend to leave a trail of bloated, duplicate code if you aren't careful.
Was dealing with this today and realized how much of a difference native LSP integration and code integration makes in Spettro.
Instead of the agent burning tool calls spamming grep and glob to find things, repo-search just indexed the codebase and functions directly.
Passed a function name, got the exact file location, line numbers, and the inefficiency, that Spettro fixed before I even committed the code.
Most agents don't have native LSP, so you end up having to hack together skills, plugins or beg the model to navigate properly.
Native LSP and code indexing inside the agent just feels like the obvious way forward.
















