vrambo8
28 posts
vrambo8
@vrambo8
AI @meta, investments n00b: building an agent that can teach me
London Katılım Ekim 2025
46 Takip Edilen71 Takipçiler
People only see the results
They don’t see the hours, the consistency and the grind behind it.
For the about 4 weeks I’ve been averaging 800 – 1,000 replies every single day spreading value and supporting other creators.
The truth is if you want results, put in the work.
Stop asking “how” and start doing..
English
Unpopular opinion:
90% of "AI startups" are just:
👉🏻 GPT wrapper
👉🏻 Nice UI
👉🏻 $10M valuation
The real AI companies are:
👉🏻 Building infrastructure (compute, data)
👉🏻 Solving unsexy problems
👉🏻 Profitable in year 1
Which camp is your startup in?
GIF
English
While everyone else is adding usage caps and tier limits, VEED goes unlimited across all plans.
This is what happens when a company actually listens to creators.
WAN 2.2 + no generation limits = content teams are about to eat.
VEED | AI Video Creation@veedstudio
VEED Motion Unlimited is LIVE 💥 This makes us the FIRST to offer unlimited video generation on all plans. Powered by WAN 2.2. Make your brand go crazy with content.. Free 500 credits for first 200 users in DMs, who: • Retweet • Comment “VEED”
English
github repo for the LLM proxy API is up: github.com/vrambo8/x402-a…
i want to build open-source because it’s the best way to earn trust and get the sharpest minds collaborating for the common good.
looking for testers and early feedback. let's go
English
@HarishKotra @GaiaNet Totally agree! Unfortunately accessibility to GPU resources and cost associated with it means that not everybody will be able to host their own models.
Hopefully costs will lower as development advances and we’ll reach true decentralized AI.
English
I would take it a step ahead and say, own your model, own your data and don't pay blackbox AIs if you truly want to build decentralized AI: x.com/Gaianet_AI/sta…
Make your own custom chargeable endpoints with your fully owned AI Agent using @GaiaNet
vrambo8@vrambo8
AI agents can use x402 to enable agentic commerce, but they still need credit cards for LLM API providers a fully autonomous agent should be able to use x402 to pay for OpenAI tokens first demo on testnet ready, stay tuned for GH repo + mainnet
English
Loving x402,
but verify-settle-execute breaks when costs only show up after execution.
I’m building an LLM API proxy to bridge x402 payments with AI and hit this wall hard.
No protocol fix yet that I know of—for now, I’m using software patterns (escrow + post reconciliation).
English
@milesdeutscher @henloitsjoyce Great overview! I’m building an x402 proxy for LLM APIs for AI agents to be fully autonomous
x.com/vrambo8/status…
English
@dabit3 Amazing explanation! Really helped get started with my x402 proxy for LLM APIs
x.com/vrambo8/status…
English
x402 explained in 100 seconds ⚡️
crypto adoption is hitting an inflection point, with x402 as a major catalyst.
in this video I break down what it is and why it's important.
with links to my favorite resources for building and getting up to speed.
English
We solved this by using permit based signatures (erc2612) instead of transfer with authorization (erc3009).
This lets you verify that a max amount has been authorized, do the inference, then take the payment according to the tokens used with a ceiling of the max amount authorized.
All you need to do to switch to permit signatures is pass "permit" as the primaryType of the asset eip712 definition (addition to the name/version prop already in the standard reference lib)
DMs open if you need help setting this up!
English
@vrambo8 Check out our escrow and aribtration primitives! We have an integration with x402 as well.
Intro video: x.com/arkhai_io/stat…
x402 integration: x.com/arkhai_io/stat…
Docs: arkhai.io/docs
Arkhai@arkhai_io
Now we're integrating x402 for instant HTTP-native payments. Soon agents will request services, receive payment requirements via HTTP 402, and settle through escrow - releasing funds only when conditions are met. Code: github.com/arkhai-io/alka… Demo: youtu.be/h6yxylz2_5Q?si…
English
@SuriPuri23 would be amazing
"while a theoretical upto would transfer up to an amount, based on the resources consumed during a request (ex: generating tokens from an LLM)."
though implementation is not there yet
English
