Building | DeFi Enthusiast | eng @GoldmanSachs | @RiceUniversity Alum | Opinions are my own
Joined April 2022
- Tweets479
- Following1.3K
- Followers173
- Likes9K
Cafes in SF block charging ports, and disable WIFI on the weekend.
Quite the strategy to keep people moving..
Mohamed Abead retweeted
Underrated life advice: Keep proof that good things happen to you. Old photos. Kind messages. Positive feedback. Notes from great days. Whatever. Store them in a single place. Call it a Rainy Day Folder. Pop it open on a bad day and watch your mood turn. It's a free reset button.
Mohamed Abead retweeted
told my european friend i was a techno optimist and he responded that he prefers house music
Muse is amazing! Only concern is data privacy, but honestly one of the best products I have used so far. Took <1 min to get up and running. Seems well integrated with Meta Suite from the get-go too!
Mohamed Abead retweeted
My favorite use cases for Muse / Instinct so far:
1. Submit FOIA requests to request data from the US government
2. Creating spend-limited Privacy cards to spend on subscriptions without having them recur
3. End to end filed an entire visa form for a country
4. Responded to a coordination mail for a wedding by finding the flight and hotel details
5. Look for reservations for restaurants or concerts when they open and purchase them immediately
A lot of the web was designed with dark patterns: increase friction to prevent enough humans from doing something, and now those walls are completely broken.
At this point, I feel like I’m squarely limited by creativity and understanding what is possible.
Mohamed Abead retweeted
Hey ████,
I think there is room for both.
1. Prototypes and other throw-away code can be treated as totally black box. If you’re going to throw it away anyway, and if the blast radius of it breaking is low, it doesn’t need to be perfect.
2. Production code written by Claude should have a higher bar than if it was written by a human. At Anthropic, we have many guardrails in place to make sure this is happening: lots of lint rules, lots of tests, Claude-driven end to end tests, Claude-powered fuzzers running daily, automated code reviews and security reviews, automated code refactoring, and so on. Without these, you can end up with a mess that is hard to maintain down the line. Luckily, the model makes it increasingly easy to do these well — run a few daily routines, use Claude Code Review, etc.
Your job is to hold the bar on code quality. If Claude’s code doesn’t meet the bar, try:
- Using the latest frontier model (Opus 5 or Fable 5.1)
- Increase effort to high or xhigh
- Invest in your CLAUDE.md and skills to succinctly teach Claude how to work in your codebase
If all else fails, steer Claude more when you work with it, or have Claude fix accumulated debt and rewrite your codebase to make it easier to work with. Or, wait for the next model.
Best,
Boris
I was talking to a friend today who works at an "AI" company that just allowed them access to coding AI-assistants last month. We are still early
Mohamed Abead retweeted
But yet, you are ignoring the actual problem set we are solving for here:
- Fully headless tool use access, using client-side signatures to authorize payments without manual signup or per-provider API-key provisioning.
- Prices and payment requirements surfaced directly in 402 responses, allowing agents to evaluate whether a resource is worth buying before paying.
- Autonomous runtime loops where agent deployments can both earn and spend: receive payment for work, then use those funds to purchase inference, data, or tools, all programmatically within the same runtime.
An agent shouldn’t need to navigate a bank’s website through browser automation to participate in commerce. A wallet gives it a native way to receive and deploy funds.
The reality here is that we are using cryptography/crypto to improve ux/context efficiency for agentic workflows, reduce tool config friction, etc via constructs like x402. If you don’t get it…I’m sorry…
Mohamed Abead retweeted
My life improved dramatically when I stopped trying to squeeze value out of every minute. Some of the best parts of life are inefficient. Wandering walks. Long dinners. Sitting around with people you love. Not every hour needs an outcome. Don’t optimize the life out of your life.
I was just thinking what happens if a client sends onchain tx after a x402 only to get a 5XX error
Solution: x402 splits it into verify step and settle step where it only settles if upstream succeeds
This is bad for the industry and innovation. Creating a hostile environment for collaboration brings down overall innovation in the space.
More importantly, AI labs with frontier models and large capital can cut down products by simply stopping their usage. It infringes on the health of the entire startup ecosystem.
We’re ending our partnership with Cursor following its acquisition by SpaceX. Under our proposal, Cursor’s direct access to our models would end on November 12.
We know that the people most affected by this decision are the developers who rely on OpenAI models in Cursor. We care about their experience in this transition and we’re ready to go above and beyond to support them.
openai.com/index/our-decisio…
