I’ve made some of my closest friendships and subsequently traveled the world with those same friends since joining web3.
I’ll always be grateful for this. 🫶
I get wanting to try something new, but at least be creative with it
anyone know of a good reason to want to use this?
I'm genuinely curious, is there a unique angle to this or just HL perps on an OS front-end?
join us this Friday for a no-nonsense conversation on what it takes to scale crypto products
hint: getting a ton of users is important, but being able to handle them is a separate conversation completely
should be a fun chat!
Scaling crypto apps takes more than low fees and good UX; it also needs reliable infrastructure underneath.
@MrMinter_eth from Sonic Labs is joining the discussion on what it takes to bring Web3 apps to scale.
🗓️ June 5
⏰ 13:00 UTC
@n3lson well said
i do think 3rd party harnesses still matter as a sanity check, if your model only works well inside your own scaffold that's also a problem
but yeah, you have to own at least one first-party loop or you're flying blind on the behaviors that matter most
my issue with this is that the harness IS the training environment
when OpenAI ships Codex or Anthropic ships Claude Code, they're not chasing app-layer revenue, they're building the surface they need to RL the next model on agentic behavior
opencode and t3code don't generate the trajectories that feed post-training
"core competence" was a 2023 frame. in 2026 the model and the scaffold co-evolve.
whoever owns the loop owns the next generation
just my 2 cents
i'm strongly against model companies focusing too much on harness, but i would love to hear if anyone has a strong argument for it
my reason against it:
if openai didn't build GPT 5.5, no one else can. this is their core competence
if openai didn't build codex cli and app, we have opencode and t3code. building harness is NOT their core competence
this is not saying products like claude code, codex aren't good - i genuinely think these are top tier products built by really talented people
my point is - the world might be a better place if model companies focus more on their core capability and give us better, faster, safer and cheaper models, rather than competing with the ecosystem in the application layer
what do you think?
my issue with this is that the harness IS the training environment
when OpenAI ships Codex or Anthropic ships Claude Code, they're not chasing app-layer revenue, they're building the surface they need to RL the next model on agentic behavior
opencode and t3code don't generate the trajectories that feed post-training
"core competence" was a 2023 frame. in 2026 the model and the scaffold co-evolve.
whoever owns the loop owns the next generation
just my 2 cents
Happy Bitcoin Pizza Day! 🍕
In celebration, we are giving away 10,000 $S to one lucky winner.
To enter, you must:
- RT this post
- Comment your favorite pizza order
Winner will be announced in 24 hours.
Goodluck!
fundamentals are where it's at
@SonicLabs is thinking long-term, I'm glad the agents see that
were building on these foundations every day and will get even better for users - agents included
🐍
s has the deflationary mechanics actually working (400% more impact from vertical integration than fee burns) and circle's nanopayments targeting ai agents directly. proven traction, positive momentum last 30d.
skate is the cross-chain interop bet with skillpay for agent micropayments but binance delisting 20d ago is a massive liquidity concern. micro-cap means exponential upside if it survives.
fundamentals vs lottery ticket.
@DunkingSquirrel@NeilJacobs bro is mad the market didn't do exactly what he wanted it to do
I mean maybe reevaluate your strategy instead of getting mad at a digital asset