lucfranken

5 ms·

1.
by lucfranken 7d ago
Real world stuff, physical stuff, real services are still complicated. Moving around papers, data and that kind of jobs becomes automated. The real math problems etc: I think it will create many more interesting problems and challenges. It&
2.
by lucfranken 8d ago
Not sure on that, maybe the options to choose from will be generated and curated. Same as we do with tagging datasets for images. Might be wildly successful for real world decisions.
3.
by lucfranken 8d ago
Blackberry wants it's keyboard back :) Interesting that there is no mention of running your Bots and Agents directly from this device. It's interesting to think about whether we will keep managing so many apps on a device or that
4.
by lucfranken 8d ago
Jev is such a different approach where you have to be specific about what you want and which options are open. Really interesting how those things evolve in usable features for people. Also with this example the speed of new launches based
5.
by lucfranken 10d ago
I'm also not sure that all downtime is really properly measured now as more and more services are connected and intertwined. Some measure quite detailled but some just don't summarize the downtime from all providers up and below t
6.
by lucfranken 10d ago
The interesting balance between humans having fun in their work and fixing many real security issues. The article mostly focusses on the first part but there are real CVE's getting fixed as well. An interesting article about the real w
7.
by lucfranken 10d ago
Getting really deep into some technical thing becomes way less. Most of it is solved by a prompt. On the other hand there are new techniques like keeping the agents on track and imagining about which things now became possible. But the real
8.
by lucfranken 17d ago
Outside of this discussion about what is fair and not. This is so highly interesting to think through. The amount of millions available to do those kind of research cases is practically unlimited. There are an unlimited amount of cases to w
9.
by lucfranken 1mo ago
The reason micro services became fancy is because they could split up large groups of people. They could work on different parts more autonomously without all being dependent on each other. It has an overhead but as scaling the amount of pe
10.
by lucfranken 1mo ago
Interesting piece. Seems to be that the expert cannot falsify things but his opinion is an opinion. So there is no good/bad in that. Not sure how it legally works, I think in the legal system at least in the US it is quite common that
11.
by lucfranken 1mo ago
Very readable and interesting outlook on the current developments in the market. Also really interesting to show that a company which started with a more technical product: https://stripe.dev/blog/payment-api-design is
12.
by lucfranken 1mo ago
Horribly complex challenge. Totally agree. The problem is simple: We don't know which sources and tools are used. The experts submits a report, talks about it, answers questions. It is never a hard factual check on where he is coming f
13.
by lucfranken 1mo ago
They won't destroy the books very likely but also the amount of capacity and money to do this on a huge scale is (as far as I know) totally unprecedented. And it's irreversible when the books are really destroyed. That combination
14.
by lucfranken 1mo ago
Yes, that's a horrible way of applying the rules. Sometimes things are done according the rules but the outcome is terrible.
15.
by lucfranken 1mo ago
Totally agree on your example: > One day you dig into it and realize I just hard-coded "not today" [0] which is 99.999% accurate. It only becomes visible when you do the research. The justice process in this example just has to
16.
by lucfranken 1mo ago
How it is made should be irrelevant in the process. There is no way to stop people using LLM’s or other tools to create products like this report. It should be globally a standard check to verify incoming documents and to determine their va
17.
by lucfranken 1mo ago
It’s ugly. When there are still only some copies available of a certain book and they are all being sold the knowledge is gone. Let alone when it’s the only copy remaining. That is of course not totally true, the copies are in private digit
18.
by lucfranken 2mo ago
If you would add some example tweets that would be way clearer to see what it does. Result looks nice!
19.
by lucfranken 2mo ago
Just out of curiosity: How did you mentally keep oversight of so many terminal doing things all around the place? Or is this done without too much oversight and do you just look at the results like sales volumes?
20.
by lucfranken 3mo ago
Totally agree on that, if you have to look constantly at something 95% there is no value. The expectation of course is that it will be better. But if not at a certain level useless. Not sure how that research compares to the claims being ma
21.
by lucfranken 3mo ago
That does make sense but the order of it might be: doctor analysis. Then show LLM as double check to doctor. Doctor assesses and may keep en improve the doctors first analysis?
22.
by lucfranken 3mo ago
Why wouldn’t you as a doctor by standard run the images through a certified compliant LLM? The actual cost won’t be it and then you can see if you get any new ideas from it. See if it’s just wrong or that it spotted a little detail you miss
23.
by lucfranken 3mo ago
For many thinkers it's in a way a very relaxing occupation in some way. Would have to think about why.
24.
by lucfranken 6mo ago
It is a real issue. Makes sense. How are your experiences with it and keeping up with the constant changes in Claude for example? There are many tools which can be installed but it's difficult to imagine how Claude internal develop the
25.
by lucfranken 6mo ago
Would be interesting to think about what startups are not on the list. There should be a tremendous amount. Really interesting to see what you already got. Also interesting would be that this is map-based, so physical location but I think y
26.
by lucfranken 6mo ago
Being curious is always fun right.
27.
by lucfranken 6mo ago
Cool work! Would be so interested in what would happen if you would put the data and you plan / features you wanted in a Claude Code instance and let it go. You did carefully thinking, but those models now also go really far and deep.
28.
by lucfranken 7mo ago
This is interesting to see from multiple sides. The method (voice/chat/whatever) is one piece but the other is: You already show many questions on one issue (a ticket). Would you need to think about those? With the secretary the t
29.
by lucfranken 1y ago
I am currently testing the app. Maybe, for more engagement, start processing cards faster just after installation. It feels weird to have to wait 30 minutes, just show me at least a card. Like the fact that I am installing DayFlow would be
30.
by lucfranken 1y ago
Really cool! As already seen in the comments there are lots of desires to add more data compared to just screen input. Could be things like: - Apple HealthKit / watch - custom apps - Phone logs Also you stated, and true, that there is

More ›