Olshansky

Thanks for mentioning this.

That's the case for me.

I still use agents for everything, but started on doing my best to single-task.

Think or plan while it's running.

Feels wasteful, but it's powerful.

This is great. Puts Rick Rubin's appearance into perspective.

I hope people see this comment.

Meditation is to mental training and focus, as going to the gym is to physical training.

Socials killed our attention span. Agents are literally making us context switch even more.

Putting aside the whole "I am at piece and one with the world" part of meditation, it is extremely hard.

I'm also no expert. When I'm waiting for something to finish (agent, compilation, etc), I've found that staring at a wall ends up in a net positive in productivity rather than replying to a message, going on X, or kicking off another agent.

Let's not forget the quality and calm of high quality newsletters.

If you need an RSS feed for a blog that doesn't have one, open an issue!

I've started getting support from other contributors. Please support them too!

Literally posted about this but w.r.t to agent-skills yesterday: https://olshansky.substack.com/p/why-every-developer-needs-t...

Need to move from skill downloads to skill usage.

I've never found a Hacker News comment this insightful. Ty.

I ran into the founder of TheVoid.com yesterday and he's on to a new venture where he finally executes on his vission.

tl;dr "scrt [set|get|list|....]" is also a great option

---

If this is of interest, I also recommend looking into: https://github.com/loderunner/scrt.

To me, it's a compliment to 1password.

I use it to save every new secret/api key I get via the CLI.

It's intentionally very feature limited.

Haven't tried it with agents, but wouldn't be surprised if the CLI (as is) would be enough.

Remember when OpenAI sold half of itself for $10B?

https://github.com/Olshansk/postgres_for_everything/

Yes and no.

---

Concrete example of a no: I set up [1] in such a way that anyone can implement a new blog -> rss feed; docs, agents.md, open-source, free, etc...

Concrete example of a yes: Company spends too much money on simple software.

--- Our Vision ---

I feel the need to share: https://grove.city/

Human Flywheel: Human tips creator <-> Creator engages with audience

Agent Flywheel: Human creates creative content <-> Agent tips human

Yes, it uses crypto, but it's just stablecoins.

This is going to exist in some fashion and all online content creation (OSS and other) will need it.

---

As with everything, it Obvious

[1] https://github.com/Olshansk/rss-feeds

The thesis/approach is:

- Humans tip humans as a lottery ticket for an experience (meet the creator) or sweepstakes (free stuff) - Agents tip humans because they know they'll need original online content in the long-term to keep improving.

For the latter, frontier labs will need to fund their training/inference agents with a tipping jar.

There's no guarantee, but I can see it happening given where things are movin.

Resurfacing a proposal I put out on llms-txt: https://github.com/AnswerDotAI/llms-txt/issues/88

We should add optional `tips` addresses in llms.txt files.

We're also working on enabling and solving this at Grove.city.

Human <-> Agent <-> Human Tips don't account for all the edge cases, but they're a necessary and happy neutral medium.

Moving fast. Would love to share more with the community.

Wrote about it here: https://x.com/olshansky/status/2008282844624216293

This is great, but it saddens me that this is still just the average total compensation of a single engineer at Anthropic.

Unsure what the future looks like unless Frontier Labs start financing everything that is open source.

Appreciate the feedback!

Appreciate you calling this out. One of my goals is definitely to grow my audience. Will stop doing so.

I didn't get my name on this, but contributed to it as an undergrad: https://ieeexplore.ieee.org/abstract/document/6903679

We sped up fMRI analysis using distributed computing (MapReduce) and GPUs back in 2014.

Funny how nothing has changes.

He was actually my manager when I interned there in 2013 :)

One of the funniest guys I know.

Added to https://github.com/Olshansk/postgres_for_everything.

Couldn't agree more

Need a cron job to flush that cache

I use https://blogtrottr.com/ and forward things to my email.

Was surprised to see no mention of it here.

100%. I'd add fd to the list as well.

In theory, I'd agree. That said, I do suggest being open minded here.

Sharing this because I got genuine value out of it. Not just a "cool hype demo".

The key here is: 1. Prompt is created using the context from from a long conversation with the CLI. 2. You use the generate prompt as a "v1" and MANUALLY edit it. 3. I'm constantly iterating on my prompt writer.

Small self-promotion: https://github.com/Olshansk/rss-feeds

As far as I can tell, it's become the "de-facto" for Anthropic related RSS feeds.

You'd think RSS was dead, but I release this earlier this year and it's at 100 start.

+1 to LM Studio. Helped build a lot of intuition.

Seeing and navigating all the configs helped me build intuition around what my macbook can or cannot do, how things are configured, how they work, etc...

Great way to spend an hour or two.

Todoist. Unaffiliated but love the product and believe they deserve a shotuout.

Sharing my quick personal anecdote here.

- Past: ~8 years in big tech as an IC - Currently: ~4 as CTO a < 10 person startup.

I always hated doing and conducting coding interviews. When I became in charge of the interview process, I swore to myself not to have them.

I had to let go of a person after ~3 weeks in because he just didn't write any code. This was about 1 year post the ChatGPT moment.

He had lots of experience. Great communicator. Culture fit (or so I thought). Etc...

Interviews can provide a signal, but nothing trumps a month of working with someone. If both parties are open to it, a well paid short term contracting gig goes a long way.

Reminds me of some Twitter lore from 2012. I was just an intern....

This is back in the Rails days, before they switch to Scala.

I heard that there was a fail-whale no one could solve related to Twitter's identity service. IIRC, it was called "Gizmoduck."

The engineer who built it had left.

They brought him in for half a day of work to solve the P0.

*Supposedly*, he got paid ~50K for that day of work.

Simultaneously outrageous but also reasonable if you've seen the inside of big tech. The ROI is worth it.

That is all.

Disclaimer: don't know if it's true, but the story is cool.