Germinating @bloom_directory | rust enjoyer 🦀 | @safe Guardian 🔰 | Retired MEVoor | "devanon" on decentralized media
Joined December 2021
- Tweets 2,514
- Following 493
- Followers 3,604
- Likes 3,815
194 Photos and videos
Pinned Tweet
May 12
We're launching Bloom (@bloom_directory): Ethereum as a filesystem.
Chain state becomes files.
Transactions become staged writes.
Agents can use ls, cat, jq, pipes, and policy checks instead of hand-rolled Web3 glue.
3
3
33
2,852
Devan Non retweeted
Jun 4
Something big is blooming this week, keep your eyes open 👀
2
4
8
259
May 28
1am in Thailand. I've just switched my Hermes Agent (Violet) to Opus 4.8. I've got Hermes Kanbans shipping code. Violet can now procreate, created two agents for colleagues in 5 mins. We're scaling @bloom_directory.
Going to get a foot massage and chill tf out
Thailandmaxxing
4
244
May 28
Honoured that @safe continues to support Eternal Safe, the decentralized gateway into your Safe{Wallet}. 💚
V2 coming soon with many QoL updates and immutable Safenet staking!
Learn more on GitHub (first reply). 👇
Meet the six teams bringing SAFE staking everywhere:
• @cobuildersxyz: a staking dashboard, a CLI, and an MCP server to stake SAFE from Claude Desktop or Cursor
• @bootnodedev: a dashboard with live validator insights and a plain-language risk preview before you sign
• @protofire: a production-grade dashboard with a 99% uptime commitment
• @devanoneth: SAFE staking built into Eternal Safe, the IPFS-hosted Safe wallet
• Chainza: a one-import widget any wallet or app can embed to offer SAFE staking
• Core Contributors: keeping today's reference staking interface live and maintained
Building in the open. Follow along.
1
2
5
530
Devan Non retweeted
May 28
New pitch for Bloom:
We’re building software nutrition labels that your agent can verify at runtime
May 27
In an ideal world all software and hardware would have "nutrition labels" that provide a full list of trust dependencies - what math and which actors' honest behavior (and on what time scale) the system is relying on to provide its core functionality and implied guarantees.
1
5
201
May 27
Happy to report I'm at stage 4 with Hermes now.
For a few days there I was reporting to my colleagues that it was a big load of carabao shit.
It's more like a big load of sane defaults for an LLM harness. Nice.
1
163
May 27
So I've got Violet (my Hermes agent) setup to a nice point. Now it's time to setup Hermes agents for my colleagues. Of course, I'm not going to do that, Violet is.
Time for her to procreate.
(wtf is this timeline)
2
1
196
May 27
Prompt:
> You are going to procreate. I want you to SSH into the Debian VPS that I am going to setup and install Hermes there. Get it setup to the point that @_apedev can SSH in to provision accounts, etc. Give me your public key and a plan.
145
May 27
Whoop is worth every cent.
How many more years of life did I gain with this improvement? How much more quality do I get every day?
Tbf I’ve been making real lifestyle changes so it’s not the Whoop alone, but as a dude that loves numbers, it certainly helps.
3
231
May 26
We need a new foundation for this new world of software.
1
5
201
May 25
I feel like learning Hermes right now is kind of like being early to Linux back in the day. Sure agents are going to get better and better at setting themselves up, but I think the need for a human to know the internals of Hermes will never go away, especially at scale.
1
4
192
May 22
I’m setting up personal agents for everyone @bloom_directory
Meet Violet: Hermes, 6GB RAM VPS, Codex and Claude, Discord, Telegram, GitHub, Google workspace account. They need to be everywhere we are.
What are some good jobs I can give her?
1
3
231
May 21
Post about exciting new technology that will reshape software for agents: 4 likes
Post a photo working from a typical van in Thailand: 19 likes
Chat, how do I invert this?
1
6
178
May 20
You missed the best part:
> Rubik’s cube YouTube tutorials (2007 - 2011)
May 19
Andrej Karpathy's incredible resume:
> Google, Working on DeepMind (2015)
> OpenAI, Founding member (2016 - 2017)
> Tesla, Senior Director of AI (2017 - 2022)
> Anthropic, Working on R&D (2026)
5
251
Devan Non retweeted
May 19
Bloom’s vision lines up very closely with the direction @VitalikButerin is describing.
In the agentic-era, the bottleneck shifts from code generation to verification, trust, and coordination.
May 18
Many people have claimed that with AI-assisted bug finding, secure code (and hence trustless anything) will be impossible.
I have a much more optimistic take, and AI-assisted formal verification is a major part of the reason why:
vitalik.eth.limo/general/202…
1
4
9
461