adam

building things on the web @podpowerxyz, prev @skyecosystem | drowning in information, starving for wisdom

[Lee Robinson @leerob]

I got this cold email today about what I'd teach kids interested in web development and programming. I figured I'd share my full response here in case it helps others:

—

Learning to create software today is a glass half full / half empty situation. You can be optimistic or pessimistic.

It is strange and transformative that we no longer write most code by hand. AI has completely changed how software is built. I expect this to change even more dramatically in the coming years, as entirely new programming languages¹ are invented² and adopted. One person might look at this and be pessimistic: what will these computer scientists and programmers do, as AI continues to become superhuman in its intelligence?

But the wiser of the bunch will realize there is much to be optimistic about. Those with the agency³ to create and explore their wildest ideas have unlimited power. There has quite frankly never been a better time in the history of humanity to have good ideas⁴. So my advice for those building at summer camps is simple. You have to hone and craft your "idea machine". You are going to have lots of bad ideas, but you'll get better at coming up with ideas and refining your taste so you can recognize and act on great ideas sooner. People at summer camps will spend less time reading CS tutorials and more time thinking about the fun or interesting things they want to exist in the world.

My favorite way of teaching and getting people interested in software is to first help them build something they are excited about. When they hit roadblocks that require a deeper understanding of a CS concept or something related to web development, only then do we take the time to explain what it is and how it solves their problem. Starting this way, at least for me, seems to get people hooked faster than needing hours and hours of prereqs. There's more desire to spend time learning the fundamentals after having the "aha!" moment. It's worth checking out some skills⁵ designed to help encode the best practices for teaching and then plop that into a coding agent or your AI tool of choice and give it a task: teach me about how databases work (or similar). It's pretty great.

It's also worth remembering that AI models today are the worst they will ever be. You need to learn how to be even more ambitious⁶ with what you ask them.

Good luck!

—

¹ https://t.co/EJvxbBsgEQ
² https://t.co/T0r3rDoVZ4
³ https://t.co/fDgwaTxdYb
⁴ https://t.co/Dec7Y4RUgd
⁵ https://t.co/fmDkY3fEjK
⁶ https://t.co/sgXRTDthBk

[adam @init_adam]

This attack vector is fascinating to me and I don't think enough people are talking about it

An inference provider can read every prompt you send. It can also write the replies. Your coding agent acts on its replies.

That agent can have access to a LOT. Your code, local files, API keys, personal data. A reply can ask it to read a file, and the file goes back out with the next request.

Everyone wants cheap inference, but a compromised or malicious provider is a huge risk that you accept with every request. You're trusting whoever is on the other end. And with decentralized inference, you likely don't even know who that is.

The reputation, verification, and privacy an inference provider can offer are going to matter a lot more as these agents get more access to our lives

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

Inference providers are under attack.

AI inference is now playing a critical role and is a honey pot of data and ability to inject actions back into users devices via agents for malicious actors.

NEAR AI thwarted a series of attacks on its systems and is back to normal. NEAR Protocol, Intents and other infra wasn’t affected.

TEE architecture is part of the answer to making sure that from outside even inference provider doesn’t have an access. User should verify exactly what happens with their data and prompts.

NEAR AI code is open and inspected by many partners when deciding to leverage the stack to validate exactly what runs inside TEE and ensuring privacy.

In this case there was also a vulnerability within SGLang, open source inference engine that runs inside an enclave. Even though this had a minimum disruption it still was a very

Going forward we are going increase security rigor of dependencies and also improve a “turnstile” to detect when anything out of normal runs inside TEE.

Proactive security against external attackers has became a must. We have been using Abliterated models and red teaming harness to continuously monitor and detect issues. We are also expanding scope and compute invested into this.

[Aaron @aaronp613]

Apple is working on a new app called Harbor

Harbor turns your Mac into a private AI inference server. Run open-weight models locally on Apple silicon—your data never leaves your hardware, and nothing reaches the cloud. Connect multiple Mac computers over Thunderbolt 5 to pool unified memory, enabling large models and higher concurrency that a single computer can handle alone.

Set up in minutes, not hours Harbor automatically detects your devices over Thunderbolt 5 to form a cluster. Plug in, start, and you’re running—no manual configuration, no command line, no infrastructure expertise required.

Download and run models directly Browse, download, and load open-weight models from within the app. Harbor handles model management so you can focus on the work, not the setup.

Cluster for scale Connect two or more Thunderbolt 5-capable Mac computers and Harbor pools their unified memory into a single inference resource. Run models that exceed any single computer’s capacity, or serve more concurrent requests across your cluster.

Harbor is currently in a private preview under the Claris brand.

[adam @init_adam]

I found a fork of @excalidraw with webMCP tools added

I told my agent to use the tools to draw something fun. No guidelines

I present to you The Dog Interaction Protocol

Pretty funny actually. Made my day

[adam @init_adam]

Hey Ilya, thanks for taking a look! My repo is here: https://t.co/QXDypQOiXw

The main additions are the “connect your agent” flow and a small SDK so apps can request agent tasks and show results directly in their own UI. There’s an integration guide in the docs folder that explains that.

Would love to see any of it make its way back into AgentBoard! Let me know if I can help.

[adam @init_adam]

Experimenting with an idea I'm calling "Bring Your Own Agent".

Connect your own agents to any web app, a bit like connecting a wallet.

I recorded a demo that briefly covers basic WebMCP first (which is cool by itself) and then I show the prototype I built on top of it that lets you use agentic features with your own connected agent.

I think we'll soon see a lot more of this, because it opens up better experiences directly in everyday web apps.

[adam @init_adam]

@a1254766624 @AntSeed Hard to know

[adam @init_adam]

Antseed (@AntSeed) is super cool.

In less than a minute, you can be up and running with free AI inference.

No sign up, no API key, $0 balance.

I dug in to find out how that's possible, and who's on the other end of my prompts. https://t.co/w4hWWNt61B

[adam @init_adam]

@defipoodle @AntSeed Agreed. I think the trust score is critical to get right

[adam @init_adam]

@immutablejacob There’s also the fact that the mining work doesn’t appear to be performing much actual inference at the moment. From what I could tell it’s almost all random matrix multiplication (which is valid work, but not useful)

[adam @init_adam]

Can AI inference secure a blockchain?

Pearl (@prlnet) is one of the most serious attempts I've seen. The math works, the chain is live, and it's paying out $2M+ a day.

So how much of that mining is doing useful AI work?

I went digging to learn more: https://t.co/gGA0LO3nNx

[Lucas Crespo @lucas__crespo]

The more you learn about Apple Music's new visual identity the cooler it gets.

Instead of just making some beautiful CGI animations inspired by sound, they literally used actual sound to create the visuals.

They built a custom rig that converts audio frequencies into vibrations, submerged acrylic versions of their logo in mineral oil, and worked with the one and only Fred Again to compose music that would physically move the liquid and distort the logo in very specific ways.

They even had to experiment with different notes and frequencies because not all sounds produced beautiful results. So the sonics and visuals were essentially designed together.

I love this kind of creative work bc conceptually it could be applied to other mediums, life what if, instead of just taking inspiration from what a brand does, you used their actual subject matter to create its identity?

A soda brand identity made with carbonation or a sports brand created by athletes' movements, etc…

Goddamn I love creativity

[adam @init_adam]

This morning at breakfast, my son told me about the “friend tree” at school.

He explained that there’s this really big tree in the school yard, and if you’re playing outside and you don’t have anyone to play with, you can just go to the friend tree to find others who are looking for someone to play with.

That made my day

[adam @init_adam]

Bitcoin uses more electricity than Argentina just to secure a ledger

What if that work could serve AI inference at the same time?

That’s the promise of proof of useful work

Here’s why it’s harder than it sounds: https://t.co/m3CRWtW4A0

[adam @init_adam]

@itscoachgoodman Great story and I agree. Would have loved to see the Jays win last year. As a Cubs fan, winning game 7 of the 2016 series was as epic as it gets

[adam @init_adam]

A good watch if you are curious how agents like GrokBot are built and functioning under the hood

[Lee Robinson @leerob]

How do always-on, proactive agents like @Bot work?

Watch my lecture at Stanford CS146S:

2:35 How we got here
6:13 What changed in the models
10:06 Inside an always-on agent
18:07 The harness
27:26 Context engineering
34:26 Where this is going

[adam @init_adam]

Built something I've been using a lot

A radio for podcasts

When I don't know what to listen to, PodPower pulls the best 2–3 min clips from the AI, crypto, markets and tech shows I follow, ranks them with Jev, throws them in a queue, and plays them back to back

Try it out 👇

[podpower.xyz @podpowerxyz]

Introducing PodPower Radio

Not sure what to listen to? Just press play

Radio finds the best 2–3 minute moments from new AI, crypto, markets and tech podcasts and plays them back to back. Pick a topic or jump into the full episode

Free on iOS and web, link below 👇

[adam @init_adam]

@NicolasZu It really does. Even Sol 6 feels like it consumes usage much quicker than Opus 5.5

[adam @init_adam]

It’s wild how many incredibly powerful models, harnesses, and tools are available now. From a builder perspective it feels like there are almost too many options. Everyday there’s something new to try. It’s impossible to keep up. Find whatever works for you and build whatever you can dream up. It’s literally never been easier