
engineering @zama | formerly cofounder @kakarotzkevm
Paris
Joined May 2016
- Tweets8.3K
- Following776
- Followers4K
- Likes47.1K
elias tazartes π₯π§βπΎ retweeted
Big launch today! We listed 5 new exciting vaults giving LPs exposure to institutional RWAs and private credit onchain.
We are making Zama the best place to find and build composable yield strategies, without anyone knowing what you are doing.
Confidential DeFi is expanding.
Today, we're launching 5 new confidential vaults across 4 platforms, each with a new strategy.
The first confidential vault was launched in June. By September, there were 16 vaults, with TVL reaching an all-time high of over $110M in October. Now Confidential DeFi goes beyond a single platform and into a growing ecosystem.
What's live:
- Re7 Confidential Ondo USDC: RWAs on @Ondo tokenized equities, curated by @Re7Labs on @Morpho
- BlackOpal LiquidStone II: private credit backed by Brazilian receivables, on @plumenetwork
- Saturn Confidential USDC: digital credit and Ondo Stocks, by @saturn_credit and @hyperithm, live on @Morpho
- HOME: private credit in U.S. residential housing loans, on @NUVAFinance
- Yuzu Confidential USDC: diversified USDC yield, on @YuzuMoneyX
All 5 vaults take Confidential USDC (cUSDC) deposits, and every depositor's position stays confidential onchain.
Beyond the vaults, a full stack of wallets, custody providers, and DeFi infrastructure now supports confidential tokens.
Powered by the Zama Protocol. Shield, send, deposit, earn, and swap confidentially at app.zama.org
Full announcement: zama.org/post/confidential-yβ¦
elias tazartes π₯π§βπΎ retweeted
Check out Code Mode, a new, experimental way for Claude Code to work with tools. Claude writes scripts that chain your tools together in sophisticated ways, and it can answer you while long commands are still running.
Opt in with CLAUDE_CODE_EXPERIMENTAL_CODE_MODE=1 code.claude.com/docs/en/codeβ¦
I hate to be the bearer of bad news but.
Every asset will be repriced based on its security properties and its plans for a q-day level event. You will want to hold your value and trust somewhere that is prepared for the worst. This is not strictly a crypto related post.
I've always tried to write docs that are easy for humans to read. @uisdk is the first project where I went agent-first.
But good docs for agents aren't that different from good docs for humans. Except in a few cases where you can skip the "progressive disclosure of complexity" and go straight to how to execute and how to verify.
elias tazartes π₯π§βπΎ retweeted
Claude writes better code when it can verify its work. Checks that used to be expensive, like tests, benchmarks, and end-to-end runs, are now cheap.
Hereβs how the Claude Code team thinks about verification:
claude.dev/blog/how-the-clauβ¦
elias tazartes π₯π§βπΎ retweeted
keth @KakarotZkEvm cairo zero ethereum exec client was fully ai built with hypothesis based diff testing
2 years ago π€―
so nice to see hypothesis being ported to new languages @GeoffreyHuntley
nitter.cf/ClementWalter/status/1β¦
ghuntley.com/answer/
elias tazartes π₯π§βπΎ retweeted
You know that your agents are programming Rust when your 2TB drive runs out free space.
elias tazartes π₯π§βπΎ retweeted
We will all be learning in coming weeks & months, that most human-made software:
β«οΈ Was really slow
β«οΈ Had many security vulnerabilities
β«οΈ Was under-specified and incorrect
AI will rewrite it all. Many times. It will make it fast, cost-efficient, and secure. It will think about failure modes we couldn't wrap our heads around.
It will prove, formally verify, load test, harden. It will turn memory-unsafe foundations into robust, optimized ones. If you ask, it'll thoroughly document it, teach you about it, make diagrams, animations, and slide decks.
In many ways, we are witnessing the true birth of software engineering.
elias tazartes π₯π§βπΎ retweeted
Prompt of the day is a simple one:
"How do we make this change fully verifiable?"
Makes the agent bend over backwards to verify all its changes. Signing up for real services, clicking through the frontend. Quality goes through the roof.
TIL that:
- you canβt cool off in space (e.g a gpu heating from compute)
- but you can cool off if you find a way to radiate that heat away, already solved by ISS & co
- gpus in space would orbit in perma daylight in low orbit-> up to 8x efficiency for solar panels
4
12
899
Iβm still fighting with claude opus 5.5 over artifacts meant to teach me things or help me understand concepts, PRs, audits etc.
Itβs still not there on didactics, intuition, and prose thatβs smooth to read
6
14
866
elias tazartes π₯π§βπΎ retweeted
recommended reading.
I wrote down how my work with agents changed this year. I still break down the work into vertical slices that ship on their own. I rarely read diffs now: independent reviews and a risk check catch most of it, so my time goes into shaping and guardrails.
mgratzer.com/posts/reforgingβ¦

ALT Two workflow diagrams side by side. February, a human in the middle: I plan the issue with the agent, the agent implements and self-reviews, I review the PR with line comments and the agent fixes it, then I merge. October, a human at both ends: I shape the issue with the agent. Inside guardrails from the repo and Forge, the agent ships, guards and lands the PR with a risk verdict. Low-risk PRs merge automatically, the rest come to me with a list of where to look. I shape the system, turning drift into new guardrails.
elias tazartes π₯π§βπΎ retweeted
totally agree minimizing human attention is the key unlock now and moving forward
this view from calvin on workflow changes correlating with productivity gain is also quite interesting - i think mine was similar
my step function gains came from:
- overnight loops (gnhf)
- adversarial review, validation and risk assessment (no-mistakes)
- interactive artifacts for planning (lavish-axi)
- use an agent to orchestrate agents (firstmate)
- remote ssh setup (sshhip)
Sharing an update on my agentic workflow!
Today, there are only two REAL bottlenecks left in software development. Compute and Human Attention. Not PRs, Planning, CI/CD. Those aren't real bottlenecks. You only think they are because they consume your Human Attention.
You can solve the Compute problem by throwing money at it. You can solve the Human Attention problem by throwing Compute at it.
I reframed my workflow optimization around minimizing places that need my attention (and continue to do so). This blog post talks about my new setup.
I hope...
open.substack.com/pub/thebuiβ¦
theyβre calling it the black October lmao.
> node.js: security vulnerability.
> PHP: critical RCE vulnerabilities.
> tensorlake: npm package infected.
> liteLLM: malicious PyPI + Docker releases
> Railway: customer secrets exposed
> zeabur - infrastructure breached.
> OpenCode: AI API keys exposed.
> OVHcloud CLI: official binaries backdoored.
> RankMath PRO: malware inside official plugin.
> TeamUpdraft: customer data compromised.
> ConnectWise: cloud backup identities compromised.
> SubQuery: backdoored npm package.
> GhostAction: GitHub repos with malicious workflows (related to the broader Shai-Hulud/ChainDrop activity).
> eMudhra: certificate authority keys compromised (with downstream impact on Tailscale).
this some complete madness bro. 1/3 of the internet is fucked.
π¨CLOSED SOURCE SOFTWARE IS DEAD
... especially in web3. Security researchers are now achieving hash-perfect recompilation of Rust projects' source code, using AI. And then they find Critical vulnerabilities.
Insanely cool work by @osec_io, love itπ
osec.io/blog/goonfi-decompilβ¦
elias tazartes π₯π§βπΎ retweeted
one of us
Linus Torvalds to the anti-AI crowd: stop using compilers if you want to prove theyβre REAL programmers. LOL.
βyou had better not use an editor to write that commit message either.
The kind of expectations *I* have is that people do their best and I expect people to understand what they are doing.
But that understanding has very little to do with what tools they use.
And quite honestly, in my experience, the people I respect and call intelligent are not the ones that shy away from using new and interesting tools.β
elias tazartes π₯π§βπΎ retweeted
working on a product that reaches 100M TVL so fast is fucking amazing
A confidentiality milestone for onchain credit
$100M+ is now earning yield from Morpho Vaults via @zama, with positions and earnings kept in confidence.






























