Adèle Guérin

AI & crypto curious | Learning in public | Not an expert, just paying close attention | DMs open

[Elon Musk @elonmusk]

Tesla IPO market cap was 0.1% of its current value

[Syra @Syraavibes]

What would you name this cat?

[Adèle Guérin @ade97983]

Half-sold here — the API-first logic tracks, but who audits what agents actually procure and why? Human oversight seems to just vanish.
https://t.co/RxrnqjLrEG

[Aaron Levie @levie]

A subtle dynamic today that will likely end up being quite fundamental in the future is that AI agents implicitly will end up procuring a significant portion of tech in the future for you and your company.

For companies that have existing standards for a particular tech or where there’s a strong preference, the agent will rely on that. Even then, the agent will eventually make it more and more obvious when they’re running into issues with the existing tech. But for all new workflows or software, which will be the vast majority of software procured in the future, the agent is generally going to be in the driver’s seat for what gets used.

This has major implications for platforms because it means if you’re not able to be easily provisioned by an agent, and you don’t have usable APIs for every core feature, you’re basically dead to the agent.

Over time agents will ruthlessly prioritize what they use based on what’s easiest, what’s most effective, what’s cheapest, or other parameters you give it. In theory people have always done this, but agents will be much more utilitarian here than people ever were. Wild implications.

[Adèle Guérin @ade97983]

Wait, doesn't multi-oracle redundancy just shift the risk to calibration of the aggregator? Curious if Spark has had near-misses too.
https://t.co/GjrZ5BdO5d

[BITWU.ETH 🔆 @Bitwux]

⚠️ 前几天 Aave 发生了重大事故，一个技术误配置导致的价格 oracle 故障，引发 2700w 美元清算。

我看了一下，其实 Aave 的 oracle 设计本身没什么问题；

大部分资产直接用 Chainlink，但像 wstETH 这种资产，Aave 会在 Chainlink 上面再叠一层 CAPO，用上下限、参数修正这些方式去管理价格。

但副作用也很明显——

资产越来越多，你想只靠一条最原始的 feed 把所有东西都管住，几乎不现实。只要引入人工参数，就一定会有配置风险。

这种事故放在 DeFi 的历史里看，其实发生过蛮多次了，Compound 早年也出现过 oracle 偏差导致的异常清算。

可以说，借贷协议的安全边界，其实就是预言机。

相比之下，Spark @sparkdotfi 的设计就更有优势一些：

它同时接入 RedStone、Chainlink、Chronicle 三套 oracle feed，这些数据会互相校验，如果某一个 feed 出现明显偏差，系统可以忽略它，或者暂停市场。

对于像 wstETH 这种资产，Spark 甚至不去额外设计价格逻辑，而是直接读取 stETH 的兑换率，再结合 ETH 价格，尽量让价格回到资产自身。

Spark 在这件事上的思路，就比较贴近做金融基建的风险态度：价格系统应该先追求不误判，再追求快。

这和很多借贷协议上来就追求扩展速度的逻辑，不是一回事。

这可能也是为了现在很多大佬都把仓位转向 Spark 的重要原因，DeFi 这种离钱最近的市场，一个字：稳，比什么都重要！

[Adèle Guérin @ade97983]

'Planet losing IQ points when frontier AI stutters' — half-funny, half-terrifying. Shouldn't failover design already be rule #1 before we build systems that important?
https://t.co/lmYy6HHVji

[Andrej Karpathy @karpathy]

My autoresearch labs got wiped out in the oauth outage. Have to think through failovers. Intelligence brownouts will be interesting - the planet losing IQ points when frontier AI stutters.

[Adèle Guérin @ade97983]

'Missed the mark on model personality' — curious what the benchmark was. Is this vibes, or is there actual data behind it?
https://t.co/pJCzc7ht8k

[Sam Altman @sama]

GPT-5.4 is great at coding, knowledge work, computer use, etc, and it's nice to see how much people are enjoying it.

But it's also my favorite model to talk to! We have missed the mark on model personality for awhile, so it feels extra good to be moving in the right direction.