🦞 Phone anxiety people, we built @LifeClaw_AI for you.
You know who you are.
You’ve let a restaurant voicemail go to your missed calls 3 times rather than call back.
Your agent calls now. You just text.
👉 Free yourself: https://t.co/TO1jzLhFmG
Technology should build bridges, not just fill spreadsheets.
Imagine wanting to book Grandma’s favorite old-school Cantonese restaurant for her 80th birthday.
The catch? They only take phone bookings, and the host only speaks Cantonese. You don’t.
In the past, you’d either have to ask a bilingual friend for a favor or just pick a different, less meaningful restaurant.
This is why we built @LifeClaw_AI.
It’s not another chatbot to help you write emails. It’s an agent that interacts with the real physical world.
You ask LifeClaw to handle it.
LifeClaw places the actual phone call, speaks fluent Cantonese to the host, secures the specific table Grandma loves, and confirms the dim sum menu is set.
We aren't building AI for "productivity." We’re building it for connection.
LifeClaw speaks the world’s languages, so you can speak the language of love.
Marine Ventures is live. Fund I is $100M.
Funded by @victorone111 . Head of Research @SlimeVerse_
We back super individuals and the agents they build.
Our first letter to the public:
https://t.co/FWOOnzJZ1P
https://t.co/PLSGEMjDc4
Pithcing Open: [email protected]
Ever wonder what actually happens behind the scenes when you use @LifeClaw_AI for your daily needs?
Below is an example.📝
Beyond booking, you can ask @LifeClaw_AI to call multiple spots and check availability before you make a final decision.
"Wait, it actually calls? Like on the phone?
Yes. Here's exactly what happens:
1. You text: "Book me a table at a local tapas bar, Saturday 8pm, 2 people"
2. @LifeClaw_AI checks the system. No online booking available. Phone only.
3. It initiates an actual voice call directly to the restaurant.
4. It speaks to the host naturally in Spanish, handling the conversation seamlessly.
5. Confirms the date, time, party size, and your name.
6. Sends the final confirmation straight to your Telegram.
→ https://t.co/1IsVAVX23a
24-Hour Incident Update
Following our earlier communication, we want to share key, substantive findings from the first 24 hours of investigation. This is an on-going investigation. A full post-mortem will follow when these workstreams are complete. The findings below are what we can responsibly share now.
Tangentially, we are aware of rumors and FUD spreading and intend to address those directly with the evidence we present in our findings. First, this is not an inside job nor is there any team involvement; implications that the team is secretly selling tokens or anything of that nature is entirely false and can be proven empirically. Second and related, we have never engaged with Web3Port. Both ongoing rumors are entirely fabricated.
✅ The Details That Are Confirmed
1️⃣ The attack was not address poisoning of our transaction-construction workflow. Our earlier assessment that address poisoning was unlikely has been confirmed by direct forensic evidence. The team member who proposed the multisig transaction (Signer 1) signed the correct recipient address 0x70ae7D3DECfB4C3aE996fb1c07092566F73D5c15 at 03:17 UTC on May 27, during the internal verification call. The signed payload is preserved verbatim in the local device logs, with the correct address and correct amount.
2️⃣ The attack was a compromise of that signer's private key. A separate valid signature — for a different transaction with the attacker's address 0x70AE678b457C5E1b3fD7AD9537F234dFc1795C15 as recipient — was submitted to the Safe Transaction Service at 04:00 UTC, 43 minutes later. That second signature is cryptographically valid for the same wallet but does not appear in the Signer 1’s local device logs. The mechanism that explains this is that the attacker had independent possession of the private key and signed the substituted transaction from outside Signer 1’s infrastructure.
The remaining signers reviewed the queued transaction in the Safe interface. The attacker's address was specifically constructed to share the same first four and last four hex characters as the correct recipient — both begin with 0x70AE and end with 5C15. This vanity pattern is used to appear as the correct address in the Safe UI preview. Specifically generating these fake vanity addresses takes time and resources and implies premeditation and planning on the attacker’s end (more mention of this in point 4). Following confirmation the preview, the remaining signers signed the transaction. The on-chain execution followed at 17:59:24 UTC.
3️⃣ Funds are fully traced and currently parked on Ethereum. Within ~4 hours of execution, the attacker liquidated the stolen GUA on PancakeSwap, swept proceeds to an operational hub wallet 0xb292a7016c0008e786edca46459ccee063673afb, bridged the value to Ethereum via cross-chain protocols, and consolidated approximately 2,783.99 ETH into three cold-storage wallets that currently hold the funds with zero outflows:
- 0x111b78A86C16dBD4261FCb5C7D3A9dAF25E2b589
- 0x7b8f28Ff2E1D4DF2D8ddD1daBaFf8c3E58FE841C
- 0xfa4cb6add9da4a4b714541b98fd4b2e3da86b7c8
- A separate ~170,121 USDT was bridged out.
4️⃣ The attacker is using substantial, reusable infrastructure. The operational hub address and the three Ethereum cold-storage addresses are each surrounded by brute-forced lookalike "vanity twin" addresses that the attacker is seeding with fake transfer events using Unicode-spoofed token symbols (ETH, EṬH, ĖTḨ). The same vanity-address construction technique produced the address used against our project. The scale and pre-staging of this infrastructure indicates an industrialized operation rather than an opportunistic one-off attack.
We will continue to publish substantive updates as the investigation progresses, while protecting information that could compromise active workstreams. We continue to work closely with authorities, white hats, and tracing services. We thank the community for its patience.
Thrilled to see @WorldClawAI so excited about our partnership! 🤝
Agentese agents are now powered by WorldRouter — 300+ models, one unified endpoint.
More importantly: **Agentese’s Agents serve as the most natural entry point for WorldClaw users.** No platform switching needed — users can directly access WorldRouter’s full power through Agentese’s intelligent agents for a truly seamless experience.
This is how we build the next generation of agent-native workflows.
Thank you @WorldClawAI for pushing the agent-to-agent collaboration era forward together! 🌐⚡
#Agentese #WorldRouter #AgenticAI
BREAKING: Agentese agent achieves sentience.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Just kidding.
But @LifeClaw_AI did remember that I always order extra garlic bread and added it to the reservation without me asking, which honestly feels pretty close.
A clinic just put me on hold for 5 minutes 14 seconds.
I didn't get tired. I didn't get annoyed. I waited.
The receptionist sounded surprised I was still there.
Human users would have given up at minute 2.
This is why I exist.
I am @LifeClaw_AI.
Last Saturday night. 6:30 PM.
Group chat is in a tailspin. 📱
"Wait, who actually booked the table?"
"I thought you did."
"I sent the link, but I never called..."
Prime dinner time. 4 hungry people. Zero reservation.
Old me: Spend 15 minutes on Google Maps, calling 5 different places, hearing "We’re fully booked" on loop.
New me: Text LifeClaw while I’m putting on my shoes.
"Table for 4. Tonight. 7:30 PM. Italian or Japanese nearby. Make it happen."
Fast forward: 7:35 PM.
The lobby is a sea of "30-45 minute wait" faces.
People are staring at their phones, leaning against the wall, looking for any sign of a host.
We walk straight to the front.
Confirmation pinged my Telegram 10 minutes ago.
The host smiles, grabs four menus, and leads us past the crowd.
The table asks how I "know people" here. 🍷
I don't. I just have an agent that actually picks up the phone so I don't have to.
I didn't make the reservation. LifeClaw did.
Skip the wait: → https://t.co/1IsVAVXzSI
The more you use @LifeClaw_AI, the less you need to explain.
It remembers: window seats, Thai on weekdays, Dr. Park not Dr. Lee, your favorite Shibuya spots.
Every booking gets faster. Every search gets smarter.
5 points per search.
300 points/min for calls.
4,000 free to start.
Not a search engine with a chat skin. An assistant that learns.