Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
mgdev
searching Neon…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
7 ms
·
31.
▲
by
mgdev
1y ago
If we zoom out far enough, and start to put more and more under the execution umbrella of AI, what we're actually describing here is... product development. You are constructing the set of context, policies, directed attention toward s
32.
▲
by
mgdev
1y ago
Easier than you'd think.
33.
▲
by
mgdev
1y ago
Hear, hear. Folks on places like HN should also realize that the broader US is not as liberal as the news and internet would make you think. The people who comment and debate online tend to be on the more extreme ends of both sides. The res
34.
▲
by
mgdev
1y ago
Visit SE Asia sometime and you'll experience a very different sentiment. Hundreds of millions of people rely on Meta to provide valuable services every day, some of them borderline essential. This is undebatable. The outsized public ha
35.
▲
by
mgdev
1y ago
There is a potential world where Meta uses AI as a vector to tap into the home. Like, literally building smart homes. Locally intelligent in ways that enable truly magical smart home experiences while preserving privacy and building trust.
36.
▲
by
mgdev
1y ago
Thinking on this more, I've realized there's an additional problem with Koomen's vision that might be even more significant than what I initially described. It's actually potentially harmful that current AI email assista
37.
▲
by
mgdev
1y ago
While Koomen makes valid points about the limitations of current AI implementations like Gmail's assistant, I think even his analysis misses a more fundamental insight: we're in a transitionary period that will eventually lead to
38.
▲
by
mgdev
1y ago
Who is using this?
39.
▲
by
mgdev
1y ago
Whatever Claude Code is doing in the client/prompting is making much better use of 3.7 than any other client I'm using that also uses 3.7. This is especially true for when you bump up against context limits; it can successfully re
40.
▲
by
mgdev
1y ago
These are great. I had my own list of takeaways [0] after doing this for a couple years, though I wouldn't go so far as calling mine factors. Like you, biggest one I didn't include but would now is to own the lowest level planni
41.
▲
by
mgdev
1y ago
Strictly worse than Claude Code presently, but I hope since it's open source that changes quickly.
42.
▲
by
mgdev
1y ago
I've had a really terrible experience with codex so far. It tries to code by issuing patches. The problem is the patches are always wrong, so they never apply. It just spins and spins and spins and spins... AND you have to press '
43.
▲
by
mgdev
1y ago
<3
44.
▲
by
mgdev
1y ago
I built MCP Commander to solve a small but growing problem I have: having to repeatedly configure new (and potentially untrustworthy) MCP servers across various tools and projects. MCP Commander can manage your configuration files across an
45.
▲
MCP Commander: Monitor, Debug, and Secure Your MCP Servers
(mcpcommander.com)
3 points
by
mgdev
1y ago
|
1 comments
46.
▲
by
mgdev
1y ago
...except mindshare. Turns out, that matters a lot.
47.
▲
by
mgdev
1y ago
He should care principally about his customers. That's it.
48.
▲
by
mgdev
1y ago
Automattic is a $7b+ company. That's the twenty-year body of work you should use to evaluate his execution and strategy.
49.
▲
by
mgdev
1y ago
> But they are not smart enough to imagine a way out of their little boxes. Do they know this is what you think of them? How effective has this attitude been in helping you understand their perspective?
50.
▲
by
mgdev
1y ago
Matt's only real problem is not owning his ambition openly. Trying to publicly argue the moral high ground was a stupid, unforced error. It didn't need to be moralized at all. Just make the changes you want to make, piss off a voc
51.
▲
by
mgdev
1y ago
If people you respect are swayed, ask yourself (or better: them) what they see that you don't. I doubt it's a matter of intelligence, so much as perspective. If no one you respect has been swayed, you should know: the other side i
52.
▲
by
mgdev
1y ago
A US-based dev costs 125/hr, on the low end. A US-based dev directing Claude Code has like 3x output. So the biz is spending 125 + AI costs, but saving 250/hr. An individual dev might feel like a superhuman compared to those not u
53.
▲
by
mgdev
1y ago
This is so good. I've been using it with Claude Code with great success. I just leave an instruction in CLAUDE.md to validate changes with Playwright. It automatically starts a dev server (wrote a little MCP server to do that), navigat
54.
▲
by
mgdev
2y ago
Yikes. Great way to ruin your brand. I've owned multiple Jeeps in the past. Never again.
55.
▲
by
mgdev
2y ago
"I may be early, but I'm not wrong."
56.
▲
by
mgdev
2y ago
"I have a saying which is: When the data and the anecdotes disagree, the anecdotes are usually right." - Jeff Bezos
57.
▲
by
mgdev
2y ago
It's described as a pyramid for a reason.
58.
▲
by
mgdev
2y ago
Absent examples of concrete impact (damage), it's arbitrary fear mongering. You can write about anything to make it sound bad, even when it's good, and vice versa. Need to focus on outcomes.
59.
▲
by
mgdev
2y ago
You're describing every ML ranking system.
60.
▲
by
mgdev
2y ago
The EU once again races toward premature regulation. Europe's tech sector will continue to wither as America and others surge ahead. You can't regulate your way to technological leadership.
More ›