wagcook retweetledi
wagcook
668 posts
wagcook
@wagcook
building agentic payment stuff @kleepay | ai maxi
Blue Earth, MN Katılım Ocak 2023
1.8K Takip Edilen7.4K Takipçiler
wagcook retweetledi
sequoia put out a blog post called "services is the new software"
look at this map of over $1T in services being replaced by AI agents
English
wagcook retweetledi
Agentic payments aren’t about faster checkout.
They’re about a new economic actor: software.
For decades payments assumed:
- Humans create intent
- Humans authenticate
- Humans carry liability
Agentic commerce breaks that:
- Intent is inferred
- Auth is delegated
- Execution is automated
Which forces a redesign of payments:
5 big shifts:
- Intent becomes infrastructure (AI agents control discovery)
- Identity becomes permissioned agents, not logins
- Liability models get rewritten
- Billing moves to usage & microtransactions
- Protocols become strategic control points
The real constraint isn’t technology.
It’s trust.
Adoption happens when agents can spend:
- Within clear limits
- With auditability
- With human override
Agentic payments aren’t a checkout feature.
They’re a battle for control of the payment control-plane:
AI layer vs wallets vs merchants vs protocols.
SO: Sam
English
wagcook retweetledi
Claude code source code has been leaked via a map file in their npm registry
> Sam Altman right now:
Chaofan Shou@Fried_rice
Claude code source code has been leaked via a map file in their npm registry! Code: …a8527898604c1bbb12468b1581d95e.r2.dev/src.zip
English
wagcook retweetledi
AI is a gift to ambitious people, because it brings within range things that previously weren't.
English
wagcook retweetledi
Software horror: litellm PyPI supply chain attack.
Simple `pip install litellm` was enough to exfiltrate SSH keys, AWS/GCP/Azure creds, Kubernetes configs, git credentials, env vars (all your API keys), shell history, crypto wallets, SSL private keys, CI/CD secrets, database passwords.
LiteLLM itself has 97 million downloads per month which is already terrible, but much worse, the contagion spreads to any project that depends on litellm. For example, if you did `pip install dspy` (which depended on litellm>=1.64.0), you'd also be pwnd. Same for any other large project that depended on litellm.
Afaict the poisoned version was up for only less than ~1 hour. The attack had a bug which led to its discovery - Callum McMahon was using an MCP plugin inside Cursor that pulled in litellm as a transitive dependency. When litellm 1.82.8 installed, their machine ran out of RAM and crashed. So if the attacker didn't vibe code this attack it could have been undetected for many days or weeks.
Supply chain attacks like this are basically the scariest thing imaginable in modern software. Every time you install any depedency you could be pulling in a poisoned package anywhere deep inside its entire depedency tree. This is especially risky with large projects that might have lots and lots of dependencies. The credentials that do get stolen in each attack can then be used to take over more accounts and compromise more packages.
Classical software engineering would have you believe that dependencies are good (we're building pyramids from bricks), but imo this has to be re-evaluated, and it's why I've been so growingly averse to them, preferring to use LLMs to "yoink" functionality when it's simple enough and possible.
Daniel Hnyk@hnykda
LiteLLM HAS BEEN COMPROMISED, DO NOT UPDATE. We just discovered that LiteLLM pypi release 1.82.8. It has been compromised, it contains litellm_init.pth with base64 encoded instructions to send all the credentials it can find to remote server + self-replicate. link below
English
wagcook retweetledi
wagcook retweetledi
Kind reminder: I'm available to pair with startups (or urgent bigco) to help get your services "agent-buyable".
It'd be an odd world in which agents can do nearly anything for us other than transact. Agents as the buyer, or trusted liaison, seems like a when, and less of an if.
English
Anthropic is shipping at a pace that feels insane today.
Wouldn’t be surprised if this is just normal for software teams in 6–12 months…
English
if AI can build anything, then the only thing that matters is what is worth building, and that question is still human.
Jane Manchun Wong@wongmjane
What’s the point of building a product, funding a startup, doing anything, or even existing, if AI will eventually be capable of doing everything?
English
@GeorgeLampro20 totally agree, campus life gives u a rare window to explore, build real relationships, and experiment with low risk before the real world fully kicks in.
English
Although I’m a college dropout,
I’m not bullish on dropping out to pursue your dreams.
I see so many people drop out with no actual trajectory or validation of what they’re working on
The “burn the boats” montra came from an army of soldiers getting to an island and burning their boats to make sure they had no other option but success. The thing is the soldiers came with training, weapons, and preparation.
You dropping out of highschool or college to pursue your startup with no traction or validation. Isn’t burning the boats it’s shooting yourself in the foot before you begin a marathon.
College is great because it’s 4 years of you being able to focus in on creating something without needing to get a job. If you can’t manage college work and starting a business, you’ll probably crack under the adversity of building something anyway.
Now if you’re making more money than you would be if you graduate with a degree then screw it, drop out and see what happens.
English
wagcook retweetledi
wagcook retweetledi
People who refuse to use AI are like farmers who refused to use tractors.
English
wagcook retweetledi
Curated agent marketplaces look a lot like AOL. Open payment protocols look a lot like HTTP. The last time this happened, HTTP won.
What will prevail in the age of agentic commerce?
English
wagcook retweetledi
wagcook retweetledi
How long until agentic payments overtake human payments?
Brian Armstrong@brian_armstrong
Very soon there are going to be more AI agents than humans making transactions. They can’t open a bank account, but they can own a crypto wallet. Think about it.
English
wagcook retweetledi
Obvious trend this year is the AI Agentic Economy x Crypto
AI Agents have permeated everywhere (Hermes Agent, OpenClaw, Ironclaw, Nemoclaw)
The next step is giving them wallets and the freedom to raise, transact and use crypto/stablecoins
Open source AI + permissionless money
Tommy@Shaughnessy119
Two of the biggest metas in AI right now, @openclaw and @NousResearch Hermes agent were created by open source AI teams Not centralized labs Fast forward and they are the fastest growing GitHub repos in recent memory but more importantly it is very easy for people around the world to spin them up and use for their own unique and personalized use cases - the deployment barrier is zero. A ton of people are using them and sticking with them Whenever we get a technology that reduces the barrier of adoption by an order of magnitude, you get a extreme cambrian explosion of innovation. It happened with the internet, happened with mobile, it happened with LLMs and now it's happening with agents It’s simple. When a billion people can try a technology you get more creativity a centralized team with a set number of people We’ve already gotten past the executive assistant/COO use case which is table stakes as people build games, simulate reality, patch critical software bugs and much more. I continue to believe these agents will be good for every sector of the world but I want to linger on crypto for a second Crypto reduces the barrier to capital formation, community formation, and entry to the permissionless economy by an order of magnitude. When we marry that with agents, which reduce the technical barrier to building by an order of magnitude, you get a match made in heaven I believe the next major leg up for both open source and crypto is open source AI agents using crypto The common pushback is agents could just use Stripe, and an entire class of agents will. But for the truly unique use cases that only an AI agent could think of and raise for (or pay for a service) I think that will happen on Permissionless crypto rails. The most creative ideas come from areas with the least shackles and I view crypto as the environment where ai agents will have max ease of use (are max Permissionless). It will start with agents sending stables on our behalf before it graduates to them creating new DeFi primitives, token econ and projects all together. My view is the majority of transaction volume on @solana by EoY 2026 will be agents vs humans. Cc @KyleSamani and @toly Nous is the leader here agent wise for SOL. Once the Hermes agent gets a wallet/crypto functionality the design space expands massively. Someone build this in please! I know it’s a bear market so it’s hard to get excited, but I am. My advice if you're a founder is to pivot and build for agents. My advice is if you're human, is to pivot and build one yourself.
English