India
Joined December 2010
- Tweets4.7K
- Following170
- Followers242
- Likes10.7K
Satyam retweeted
Nearly every phone and computer on Earth relies on one plain-text file to know what time it is.
When a country changes its clocks with a few weeks’ notice, or Morocco pauses daylight saving for Ramadan, someone has to update that file. Since 02005, that someone has been Paul Eggert, a computer science professor at UCLA, who does it as a volunteer.
Take a look at this great data visualization to see the full story:
na2.hubs.ly/H08qtQC0
Docker fumbled bad with agent sandboxes. They should just ship some variant of docker compose for agents. Because ultimately most of stuff people run are multi-faceted. They have dependencies and required infra to run.
Imagine you're shipping a new feature using your agents. When you're running it all in your dev box, it works because you have the whole setup ready for the agent to act on.
Now compare that with sandboxes. The promise of epehemeral nodes are great, but doesn't solve distributed setups.
The holistic way to solve this would be to create a platform that lets you have a recreatable setup of your stack. Sort of like pre-prod / stage environment for agents to interact with. And then, make that environment reproducible and ephemeral. I would pay for this setup literally.
Coding with remote agents is great until you can't completely produce an env that lets the agents test the feature.
sure about the "basically nothing"? lmao
> Be Vinod Khosla
> Born in India, move to US at 21 with basically nothing
> Co-found Sun Microsystems at 27
> Join Kleiner Perkins Put $275k into Juniper, it returns $7B
> Start Khosla Ventures, invest in OpenAI before ChatGPT existed
> Also early investor in Doordash, Affirm, Instacart, Square
> Get Ranked as the #1 VC FIrm in the world in 2023 by founders
> Buy The Seattle Seahawks for $9.6B
> Launch Reservoir
our team completely removed human reviews like two months ago and our sprint deliverables have noticebly improved now. I'm seeing so many new demos being presented every week for new features.
Satyam retweeted
> We also have traditional alerts for all of our systems, but our superfans [Claude Tag instances] can detect subtle and novel issues that make these alerts seem one-dimensional by comparison.
New on our Engineering blog: Creating in-house source control.
In May, we merged around 2,500 code changes per day. In September, we merged more than 30,000 per day. As a result, we became bottlenecked by developer infrastructure built for human (rather than agent) scale.
Here, we share how we built a new version control system—and fully migrated our monorepo from git—in five months: anthropic.com/engineering/re…
Wow what a read.
Anthropic got tired of git and created a whole suite of systems for source control to tackle agentic development within 5 months.
Git was meant for distributed offline software development.
As more orgs start spending big on agents for coding, the SVN scene that we currently have (github, gitlab, bitbucket et al) would need to revamp to support always-online agents.
New on our Engineering blog: Creating in-house source control.
In May, we merged around 2,500 code changes per day. In September, we merged more than 30,000 per day. As a result, we became bottlenecked by developer infrastructure built for human (rather than agent) scale.
Here, we share how we built a new version control system—and fully migrated our monorepo from git—in five months: anthropic.com/engineering/re…
models are now too frontier now!
you don't need skills for every single thing now. just common-sense practical prompting.
Some mistakes you might be making:
Your effort level is too high
Your skills are too long
You're not aligning before you start work
You're aligning too often, and too hard
Your automated checks are too lax
Your custom tooling is too token-inefficient
You aren't doing enough automated review
You aren't reviewing previous coding sessions (with an agent)
You're doing too much human review
You aren't vibe coding enough prototypes
Your agent can't verify its own work
You're babysitting the agent too much
You've overloaded your global steering files
Your codebase is hard to navigate
You aren't building a shared language
Your codebase architecture allows bad tests to be written
You aren't thinking strategically enough
You aren't thinking enough
Legend.
Openclaw is as significant as Linux. It opened a new paradigm of personal assistants.
Muse, Dots, Grok bot etc are just clones like Windows.
Hermes feels like Apple though
we do not give enough credit go @steipete for his work on a very ground breaking agentic framework that is not utilized by a ton of major new products.
Satyam retweeted
Hot take: I don't like queueing messages.
Modern models are smart enough to take a "steer" message and not get distracted by it