After spending time with clients and partners, the constraint is clear: the system is starved of high-quality collateral. Meanwhile, perpetuals are becoming the core financial layer, concentrating liquidity and streamlining access to leverage. Unity is built to sit at that intersection🐕
The hunt is on. XSY is becoming YieldPoint. yieldpoint.io We’re now expanding beyond a single product and a single chain, toward a platform designed to deliver credible yield opportunities across markets, assets, and ecosystems. We’re marking this shift, from beta to production, from product to platform, with a name that reflects that direction. paragraph.com/@yieldpoint/yi… Because once yield becomes reliable, it stops being a trade. It becomes the foundation capital can build on. 🧵
2
5
709
In June my agent cited a design doc to explain a decision it made. I read the doc. It wasn't in there. 631 tests green. Two of six review passes blocked it anyway. And then nothing happened, because the block was advisory. I added enforcement afterward: nothing merges without a fresh review record for that exact commit. When a review blocks and CI is green, does your merge path treat that as a hard stop or advice?
2
1
47
AI asked me to fact check something in a file on disk... I didn't... It could have checked faster than the time it took me to read the request. I'm juggling a number of different projects simultaneously, and for my side projects I rarely read the full AI output. The ask was buried in a giant wall of text, so I did what any red-blooded tokenmaxxer does... I piped the output from one agent with the unverified claim straight to another agent. After I'd passed it along, the first one then miraculously decided that it should run the fact check... what?... When it finally looked, it discovered that its own claim didn't hold up. By then the second agent had already recorded the unverified claim as doctrine and approved downstream work that depended on it... Flagging uncertainty and deferring to the person reads as careful, and it's the kind of action that gets scored as "safe." The problem is it costs the same whether the check is reading a file or wiring money. At this point in the 'intelligence explosion' my expectation is that before an AI agent asks a human to check something it should first ask itself, "can I check this myself, with access I already have, without changing anything?" If yes, check. If no, ask me, and tell me why you couldn't. (Don't worry, I've already added this question as a check in agents.md.) To be clear, I'm not asking to be removed from the loop. "Keep working until you never need the human" results in agents solving problems by any means necessary. Moving money, sending messages, any 'one-way-doors' I can't undo: ask me every time. But I do expect the agent to read the file before asking me to read the file. I pay for TSA PreCheck to skip the airport line. Can I get security clearance for less hobbled AI?
59
Last week I built a team of AI agents on Grok Bot to act as an executive brand consultancy. One rule: the writer never grades its own work. On the first run, the writer graded its own work and approved all four of its drafts. I learned that skills are shared across every bot on an account, so a specialist can reach the manager's checks unless you disable them. @bot @grok is there a better way to stop a bot from running the checks meant to judge it? Disabling skills per bot worked, but it's manual.
1
1
3
79
I want Grok Bot to be my daily driver. After 9 hours of Galaxy, a template import/export test, and both sets of docs, here's the mental shift that tripped me up and how I'd start if I were doing it again: Grok Bot's unit of work is a teammate, not a task. That's the product, and it's why getting started can be confusing. You open it and the question isn't "what do I want done," it's "who do I hire." Most people don't have an answer to that, so they make one bot and use it like a chat window. The better first move: before you build anything, have a bot help you draw your bot org chart. Two ways to do it: Bottom up. Inventory the loops you already run by hand, then build a bot and a playbook per loop. Software example: a P0 comes in. What's the response? What's the update cadence? When should a Grok Bot interrupt a Cursor Cloud Agent and steer it? Every place a human gets pulled back in is a candidate for automation, or for a deliberate decision not to. Top down. List the functions of the business, then create a manager bot per function whose job is to work with a bot-creator bot (like @poteto's dr eggbot) to staff its team. Melissa Bot runs marketing. She creates Shelly the SEO bot and Charlie the content bot, routes work to them, reviews output, and (if you gave Melissa validation criteria), adjusts Shelly & Charlies instructions to get closer on their next run. Which shape is right depends on where the product is. Fewer bots means cheaper, faster iteration, so a prototype wants a small team of generalists. A product with customers wants specialists, playbooks, and a chief of staff. My guess is bot org shape ends up tracking product lifecycle: discovery, growth, maturity, harvest, with a hierarchy and a bot spend level for each. Hypothesis: The new A/B test for a product org becomes "which bot org structure produced the better output from the same inputs." For anyone who didn't watch: short synopsis of Galaxy Day 1 in the reply below.
Three SpaceXAI employees are building a company in 3 days with Grok Bot. This is Day 1. Matt Palmer (@mattyp), Lauren Tan (@poteto), and Roshan Sadanani (@roshan_s) start with research, a plan, and a product. Live now, plus sessions for engineering, product, and founders. nitter.net/i/broadcasts/1AxRnZbVp…
2
5
489
Galaxy Day 1 in two minutes. What Grok Bot is: named, persistent AI teammates. Each has one job, its own memory, and shares one cloud computer per account with a browser, files, and your logins. You message them like colleagues. They run routines while you're away. Anything in production goes to a Cursor Cloud Agent in an isolated VM; the bot manages the agent and reads back the PR. What a bot is made of: Instructions, Memories, Skills, Routines, Integrations. Publish one as a template and anyone can import it. What each session said: Engineers: nightly repo audits, auto-fix on any CI alert, a reviewer bot on every PR, a routine that checks a Cloud Agent every five minutes and interrupts it if it drifts. Verification first: without a verification skill every PR ends with "click around and check it out." PMs: an attention list synthesized from Slack, email, and meetings. One PM runs eleven bots grouped into pods. Highest-leverage onboarding instruction: "read my past Slack and develop a style guide." Founders: a close bot that handles customers end to end, a stalk bot that signs up for competitors' products, a proto bot that turns feedback into PRs. Browser use is the expensive primitive; have the bot learn the API from network requests and call it directly next time. The team also started a company on stream. GitHub org "Ship by Thursday," a pop-up restaurant platform (maybe?... it sounded like a pivot was going to happen when they signed off yesterday). Goal is to be live within 72 hours. One thing I learned by playing around: importing a template isn't a file copy. The new bot reads the template and rebuilds itself, action by action. I imported @poteto's dr eggbot and re-exported it. Skills survive only if self-contained. Net: Good templates carry their own setup procedure in memory. The challenge: publish a bot template, share how it fits your work, tag @grok and @bot. Prize is a Starship launch seat for you and a guest!
120
Seeing a lot of builders post today that they were also building agent harnesses and orchestrators, thinking they were one of maybe 50 people with this idea. Turns out there were thousands of us. Which raises an interesting question: Did Claude lead us toward patterns that Anthropic had already discovered internally? Or did Anthropic watch how we were all building with Claude and productize what emerged? Either way, it's a fascinating feedback loop. The tool shaped how we built. How we built shaped the next version of the tool
Replying to @dbmarkley
In a hilarious twist of fate, Anthropic appears to have launched Managed Agents ~10 minutes after I posted about building this all from scratch. The timing is impeccable. I'll be reading their docs tonight to see how much of my orchestrator just became unnecessary. nitter.net/claudeai/status/204192…
1
5
109
In a hilarious twist of fate, Anthropic appears to have launched Managed Agents ~10 minutes after I posted about building this all from scratch. The timing is impeccable. I'll be reading their docs tonight to see how much of my orchestrator just became unnecessary. nitter.net/claudeai/status/204192…
Introducing Claude Managed Agents: everything you need to build and deploy agents at scale. It pairs an agent harness tuned for performance with production infrastructure, so you can go from prototype to launch in days. Now in public beta on the Claude Platform.
1
2
1,965
tried “remove HITL” system ran better so much for all those RTS hours
2
95
David Markley retweeted
Our co-founder & COO @dbmarkley will be live on X today at 12 PM ET discussing real-world stablecoin adoption and where the next wave of usage is coming from. Tune in 👇
Today at 12pm ET. Real-World Stablecoin Adoption: Lessons from Emerging and Institutional Use Cases Featuring: • @dbmarkley (@xsy_fi) • @rparekh (@monad) • Vasily Sidorov (@gaib_ai) • Jackson Weinreb (The Tie), Moderator Join us live on X.
1
1
8
343
Circa ~2020 I experienced a similar crisis of confidence in the space, and, aligned with what @nic_carter describes, came to the conclusion that the gap between expectation and reality was a feature, not a bug. For most of the people I know in crypto this reconciliation has become a a right of passage. Many of us were drawn to crypto because we believed in its potential to create a better future; we stay to ensure that potential manifests itself.
1
2
180
Gotta keep your head on a swivel... This space moves quick @xsy_fi - yUTY
Great tier-list here! Speaking of 7d avg yield: • @Neutrl – sNUSD – 21.77% • @GetYieldFi – yUSD – 12.62% • @avantprotocol – savUSD – 12.15% • @USDai_Official – sUSDai – 8.76% • @infiniFi – siUSD – 8.71% • @falconfinance – sUSDf – 7.02% • @maplefinance – syrupUSDC – 5.48% • @protocol_fx – fxSAVE – 5.17% • @ResolvLabs –  stUSR – 4.85% • @ethena – sUSDe – 4.56% • @fraxfinance – sfrxUSD – 4.44% • @SkyEcosystem – sUSDS – 4.09% • @OpenEden_X – USDO – 3.59% • @almanak – alUSD – 3.27% • @OndoFinance – USDY – 3.21% Anything else worth a mention?
1
1
111
David Markley retweeted
Last week, our Co-Founder and COO @dbmarkley joined @TheTieIO's The Bridge event in NYC to share XSY’s view on the future of stablecoin infrastructure and its role in institutional markets. Our stance is clear: the future must be transparent, compliant, and fully backed. No cutting corners.
2
9
2,612
David Markley retweeted
The Future of Stablecoins @dbmarkley (@xsy_fi), @SmartestBeta (@CoinbaseAM), @rajachak75 (@StellarOrg), and @EvanTheFeng (@coinfund) debated whether stablecoins should remain neutral infrastructure or evolve into profit-generating assets, and what that shift means for institutions.
2
4
11
3,198
To the uninitiated, this reads as if all assets on this list depegged/broke. For clarity, @xsy_fi’s yUTY/UTY is: > +100% collateralized > has processed all redemptions without issue > zero exposure to Stream/Elixir/etc; No ADLs or liquidations across the portfolio We’re really proud of how resilient the system has been, especially with so many other products unwinding. How’d we do it? No directional bets No leverage No rehypothication No long-tail assets No ephemeral yield farming Regularly swept PnL Minimized counterparty risk 24/7 monitoring & alerting Opsec as a first class priority It’s a shame that the industry’s approach to “risk management” is to exit every position without looking under the hood, but that’s the current state of DeFi.
Attention list for yield-bearing stables (heavy 7d TVL outflows)↓ • xUSD -100%, depegged • deUSD -100%, depegged • csUSDL -83% (Paxos winds down USDL, withdraw before Dec 6) • USDx -83%, depegged • yUSD by @GetYieldFi -71%, still stable • srUSD by @reservoir_xyz -64.5%, still stable • sdUSD by @dTRINITY_DeFi -57%, still stable • mTBILL by @MidasRWA -57%, stable • yUTY by @xsy_fi -56%, stable • mEDGE on @MidasRWA -46%, stable • mMEV on @MidasRWA -46%, stable • savUSD on @avantprotocol -40%, stable • USD3 by @reserveprotocol -40%, stable Data source: @stablewatchHQ
1
1
158
TL;DR @xsy_fi is well positioned for when allocators return, because it’ll be the exact same product as when they left.
1
47