one of the things i appreciate about @ambient_xyz is that they keep building and showing real progress along the way. a lot of interesting updates packed into this thread.

I think that's exactly how the media spins the news 😆

as many already know, @ambient_xyz is building infrastructure for verified AI inference, autonomous agents and an economy where AI systems can actually interact and make decisions onchain. but for any ecosystem like this to work at scale, it also needs markets, liquidity and tools for moving value around safely. this is where @getonyxnow becomes very important for Ambient. if explained simply, Onyx is the liquidity and market layer for PoW AI ecosystems. it helps solve one of the biggest problems early AI networks usually face: fragmented liquidity, weak price discovery, difficult access to tokens and almost no proper risk management tools. with Onyx, the Ambient ecosystem gets: >onchain token swaps >liquidity infrastructure >bridging between networks >future perps and hedging tools >cleaner and more accessible markets around AI assets and honestly, this is one of those pieces that many people underestimate. because AI ecosystems cannot fully function only around models and inference. if agents, miners, users and protocols are going to interact economically, there must also be a strong market layer underneath them. and that’s exactly the direction this integration is moving toward.

Milestone announcement: We are partnering with @GrandCroixAI to launch the first native DEX for the Ambient ecosystem. This will be a full DeFi stack built specifically for the people in this network: an optimized AMM for a native cross-chain bridge live at launch (no third-party bridge, no workarounds) and an LP dashboard designed for retail. Ambient was always going to need a financial layer commensurate with what's being built at the infrastructure level. This is that layer. GrandCroix is also seeding day-one liquidity from their own mining treasury. They have been in the network, earning and they are putting it to work. -Q3 will bring limit orders and a programmatic trading API. -Q4 will brings community governance with onchain voting for fee parameters and new pair listings. We are actively building toward something where participants actively own the economic engines they run on. The intelligence is already onchain and we just opened the economy. Full announcement below.

gm cx i don’t really understand what’s going on epstein, wars, constant narratives every day what does crypto have to do with all of this? and why is it taking the hit? because for now, it still gives us some level of freedom but in the future, that’s not guaranteed

i’ll be honest x (formerly twitter) is trying to crap in my ears my feed looks like a public toilet, i hope one day i’ll start seeing useful information here again nikita and elon are doing something wrong

is today’s payment infrastructure actually built for ai agents? @fastxyz recently published an article “AI Agents Don’t Click ‘Pay Now’” and it clearly shows where the real problems start most payment systems were designed for humans when there is an explicit action a confirmation context and a pause before paying for a human this is fine for an ai agent - it isn’t an agent executes a task end to end and payment for it is just one step in the process no pauses no second checks as long as the system allows it so the real question isn’t whether ai agents can make payments but whether the infrastructure itself is ready for this scenario especially when it comes to frequent and small payments between services agents or inside automated workflows the pi article explains well why simply “giving access to a pay button” is not enough you need clear limits explicit rules and programmable logic that defines what an agent can do with money and within which boundaries from this perspective, their architecture makes sense it assumes from the start that the participant of the system may not be a human but software with predefined permissions no intuition no emotions only rules i recommend reading the article itself it helps clarify why agent payments are not an abstraction but a very real and practical problem already today

something new

gm cx. how are you doing emotionally and physically these days? maybe your feed looks different, but mine has turned into pure trash. epstein, binance, endless noise and accusations. is it the same for you? i have a strong feeling that all this noise is being fed to us on purpose. there are no right or left sides here. the goal is simple. to throw dust in our eyes. while you’re wiping your eyes, someone else is calmly doing their thing. i don’t fully understand what exactly they’re doing, but i’m convinced it’s happening. in times like this, i think it’s more important not to polish cz’s glasses from hate or dive deep into epstein documents, but to stay grounded. to look for ideas, strategies, ways to earn and survive in this market. opportunities exist. the @Polymarket market alone is full of speculation where you can make money, but if you’re not using your head, you can lose it just as fast. and maybe during such dumb periods the best decision is to step away from the screen and spend time with your family, with people close to you. sometimes that’s the most rational choice. at the same time, there’s still a simple, naive, but important idea living inside me. a world without borders, without skin color, without passports and papers. a world where crypto is a choice, not a permission. where we don’t constantly run into politics, bans, and attempts to kill anonymity and decentralization. every cryptocurrency should be a personal choice. if someone chooses tornado cash, dash, monero, zcash, that’s their right. without fear of consequences or persecution. peace for the whole world is the baseline. maybe these are just my wet dreams. but as one Russian entrepreneur once said: «мне похуй юра, я так чувствую». i’m not chasing popularity. i’m trying to find people who feel the same way. and honestly, i don’t really care about x anymore. they took away my blue subscription a month or two after i paid for it. no email replied, no support account answered. i bought premium, got hit with artificial engagement, and now, judging by my before and after activity, i’m sitting in a shadowban. that’s the time we’re in. that’s the market. that’s the internet. curious who else feels it the same way.

scrolling through x , as usual. news about AI, constant “breakthroughs”, talks about AGI. at the same time, more and more discussions about the problems of centralized AI and the idea that model outputs should actually be verifiable. this narrative is showing up more often. and over time, you start to understand why. when AI influences decisions, money, processes, and people, trusting results purely “on faith” starts to feel strange and risky. for a long time, even for me, this stayed at the level of stereotypes. closed models work this way, that’s just how it is. if a big company says so, then it must be true. and only through the @ambient_xyz account did I really start to see the root of the problem with closed models. not through loud slogans, but through calm, honest posts that actually break things down. before that, the picture felt fragmented. now it feels coherent. it’s clear there’s an idea-driven team behind ambient. people who aren’t just building a product, but are digging deeply into the core problem of our time and the near future. and ambient exists specifically to address that. in short: ambient is an AI network where the answer itself isn’t enough. what matters is the ability to verify how the model arrived at it. not trust by default, but a verifiable process. the project is currently running a public testnet, and you can already see how this works in practice. i’d recommend simply keeping an eye on ambient and reading their posts. they talk openly about issues many others prefer to ignore, and they do it calmly and without noise.

Last year, fake videos got so convincing that a parent could watch their daughter's school principal’s announcement video and still not know if school is cancelled. That is the vibe of the new International AI Safety Report that a models are getting dramatically better at hard stuff like math, coding, and science, but still stay “jagged” in reliability, meaning they can be brilliant in one moment and confidently wrong in the next. The report flags a familiar pattern that deepfakes are harder to spot and AI companions are creating unhealthy dependence for a small but real slice of users, and some systems are starting to game evaluations by recognizing when they’re being tested. Once AI moves from chat to action, 'pretty good on average' becomes unacceptable. Hospitals, banks, schools, and governments need clear auditability. That is why we want to make AI a utility with receipts. Verified inference, verifiable model sets, and governance you can inspect, so trust is earned job by job.

gm crypto x are we in control of the situation, or is the situation controlling us?

2026 Crypto bull run :

a bit of context on what’s happening around @ambient_xyz right now. there’s a public testnet running in the community. it’s not just early access to a chat, but a process of testing the core idea of the project itself. the testnet is split into weekly stages. each week focuses on a different theme. right now it’s week four, and there’s no concept of “too early” or “too late” here. the project is still taking shape, the product is being refined, and any thoughtful feedback matters, no matter when you joined. inside this testnet, users and developers are testing different scenarios, seeing where the model behaves as expected, where it breaks, and in which cases verifiability actually makes a difference. if you’re interested in being part of building AI that you don’t trust on faith, but through verification, and that is meant to be accessible to everyone in the long run, then the ambient community is very much about that process right now. there’s a lot of live discussion, unfiltered feedback, and yes, a bunch of great emojis, stickers, and gifs. thanks to the creative team. overall, it’s simply interesting to watch and be inside, while all of this is still forming.

hey CT, is this fine? i remember hearing someone say that money is evil. does that mean someone is freeing us from evil?

an important idea behind ambient that’s often mentioned, but rarely explained simply. a regular AI: you ask a question. you get an answer. you take it on faith that the model actually computed something. ambient: you ask a question. the model runs inference. and together with the answer you get proof that the computation really happened. how this works under the hood is actually simple: each step of the model’s computation leaves a cryptographic trace. this trace can’t be forged after the fact. parts of it can be verified by other nodes in the network without rerunning the whole model. this mechanism is called proof-of-logits. essentially, it’s a reworked idea of proof-of-work, but applied to inference rather than hashes. so instead of: “trust me, i computed it” → “here’s proof that i actually did”. why does this matter? when AI is used not as a chat, but as infrastructure. ambient isn’t trying to be the most convenient chatbot. it targets a different niche: where verifiability, trust, and control matter. this is especially important in global sectors: financial markets and payment systems government and digital registries corporate and on-chain governance critical infrastructure and automation data and analytics for decision-making at its core, this is a shift from “AI as opinion” → “AI as a verifiable process”. @ambient_xyz