ITSOL

We build digital products that actually ship. AI, backend, infra. Warsaw. You talk to the people who write the code.

[ITSOL @ITSOL_dev]

@OpenAI Sol became our default on day one.
Better keep up, Claude.

[ITSOL @ITSOL_dev]

@OpenAI Nice

[ℏεsam @Hesamation]

Fable 5 isn't nerfed, it's SLAUGHTERED.

the problem isn't even the model itself, but the hard guardrails Anthropic has set in place.

[ITSOL @ITSOL_dev]

flattered you named the frontier model after us.

we'll take it from here.

[OpenAI @OpenAI]

Sol is our new flagship and a step function better than GPT-5.5.

Terra delivers performance competitive to GPT-5.5 at 2x lower cost.

Luna is our most cost-efficient model, delivering strong capability at our lowest cost.

Together, the GPT-5.6 family gives people and developers more choice in how they balance intelligence, speed, and cost.

[ClaudeDevs @ClaudeDevs]

Claude Tag is the next evolution of agents.

It's a proactive, multiplayer agent with memory and identity, built on top of Claude Code.

Learn more about how Claude Tag works and best practices for using it in this deep dive.

[ITSOL @ITSOL_dev]

@ClaudeDevs

[ClaudeDevs @ClaudeDevs]

The Claude Code team has been shipping with Claude Tag internally all year.

It now writes 65% of our product team's code, including most of what built Claude Tag itself.

Here are a few ways we use it every day: 🧵

https://t.co/7PLrW06TvH

[ITSOL @ITSOL_dev]

https://t.co/w1NxcV3E2N

[ITSOL @ITSOL_dev]

"Just give us 3 devs."
The most expensive sentence in software.

We turned down a six-figure project over it last month. Here's why staff augmentation quietly kills products 🧵

[ITSOL @ITSOL_dev]

What's your test for telling a vendor from a partner?

(we build our own stuff too - Canopy, free + source-available, because we run 10+ AI coding agents a day 👇)

[Damian Plewa @nixuuu]

working with LLMs is lossy compression.

the bigger the task, the more gets dropped - edge cases, project conventions, implicit constraints. you get something that looks right and misses everything you'd have caught yourself.

early on i'd write a 500-word prompt and spend hours polishing the output. the model wasn't bad. i was handing it more than it could decompress.

the real skill is knowing where to cut. which boundaries make sub-tasks self-contained. which pieces carry too much context to delegate at all.

one 2000-line prompt means a day of fixing. twenty 100-line prompts means shipping by lunch. same model, different compression ratio.

experience with LLMs isn't prompt engineering. it's task decomposition

[Damian Plewa @nixuuu]

me and multiagents

[ITSOL @ITSOL_dev]

When your developer gets tired of tab chaos and builds a whole workstation to fix it 👀 Open source, free, no strings - attached

[Damian Plewa @nixuuu]

We run Claude Code on 10+ branches every day across multiple projects.

At some point we realized we were spending more time managing terminal tabs and browser windows than reviewing what the agents actually produced.

So we built something to fix that.
https://t.co/Z9TjniU3wt