Dan Loewenherz (@dwlz) - Twitter Profili | Zamantika Mersobahis Locabet

Sabitlenmiş Tweet

I'm squarely in the camp of believing that there is an insatiable demand for software, and until that stops, software engineers are going to be in overwhelmingly short supply in the long term. AI is forcing companies to recalibrate their resources (we're all seeing that now with layoffs and freezes), but once AI is more-or-less saturated among most companies, hiring more engineers will, once again, be the only way to build more product than the competition. And yes, being better than the competition is all that matters. Not being better than the competition 2 years ago. The pre-AI world is gone and not coming back. Readjust your expectations. Companies that think they'll be able to last with tiny headcounts are going to get crushed by the companies who choose to grow and force AI adoption across large teams. Yes, there's a brief period where adoption hasn't fully hit yet, but once it has, the low water mark for productivity will adjust and you still need more people to build more product, unless you have some secret AI sauce that no one else has access to (indie solopreneur building saas products, that's not you!).

@levelsio@levelsio

Realistically I think most tech jobs are a thing of the past with AI to be honest Right now we're seeing hiring stops everywhere, next will be layoffs I think What will remain is tech founders who build things with entire AI teams which @shl says too and I believe he's right You'd be better off going into construction and building IRL things than coding right now UNLESS you want to start your own tech business where you use AI to build it I think corporations will need 10x to 100x less devs than they have now once they're augmented by AI No idea about timeline and nobody can predict but within next 5 years I think this will slowly happen and again I'm seeing it from Remote OK inb4 "no it's just your site that has a drop bro" ye ye

English

99

102

2.3K

695.8K

Dan Loewenherz@dwlz·6h

@andyfang api when?

English

0

3

172

Andy Fang@andyfang·11h

Introducing Dasher Tasks Dashers can now get paid to do general tasks. We think this will be huge for building the frontier of physical intelligence. Look forward to seeing where this goes!

English

261

144

2.9K

1.2M

Dan Loewenherz@dwlz·1d

It's only been a few weeks since I posted this, but I'm noticing more and more people are starting to come around to the idea. The only reason the impact of large, unmaintable codebases hasn't been fully realized is that the tools that generate them have been around for such a short period of time. Agentic coding is not bad. In fact, I think it's very much a net good. I think the problem is when one applies this tool to every situation without consideration. Sometimes it's not going to be the right tool, and even if it is the right tool, there are big tradeoffs to using it. The big problem right now is that not many people are thinking about those tradeoffs.

Dan Loewenherz@dwlz

After people get a few months of seeing the impact of AI-induced slop in their codebases, and the resulting slowdown this incurs, I predict we're going to see a mass reversion back to "handwritten" code. TBH I'm feeling it myself some days. I'm actually faster with Cursor tab and just nailing what needs to happen. AI has a terrible habit of spending time on things that just don't matter. Might be more "work", but in terms of clock time, I'm getting things done more quickly without involving token inference. Maybe I'm weird. Maybe this is a terrible prediction (as things frequently are with things that change so quickly), but in this case I've been observing my behavior and others for months and I'm seeing a slow steady trickle back to the "old ways".

English

3

0

15

2.1K

Dan Loewenherz@dwlz·2d

@zeeg 100% agree

Dan Loewenherz@dwlz

After people get a few months of seeing the impact of AI-induced slop in their codebases, and the resulting slowdown this incurs, I predict we're going to see a mass reversion back to "handwritten" code. TBH I'm feeling it myself some days. I'm actually faster with Cursor tab and just nailing what needs to happen. AI has a terrible habit of spending time on things that just don't matter. Might be more "work", but in terms of clock time, I'm getting things done more quickly without involving token inference. Maybe I'm weird. Maybe this is a terrible prediction (as things frequently are with things that change so quickly), but in this case I've been observing my behavior and others for months and I'm seeing a slow steady trickle back to the "old ways".

English

0

1

57

David Cramer@zeeg·3d

im fully convinced that LLMs are not an actual net productivity boost (today) they remove the barrier to get started, but they create increasingly complex software which does not appear to be maintainable so far, in my situations, they appear to slow down long term velocity

English

467

227

3.5K

656.7K

Dan Loewenherz@dwlz·3d

@sumanthvepa Wife updates count! Even if they’re not fun.

English

0

1

9

Sumanth Vepa@sumanthvepa·4d

@dwlz I’m bootstrapped, investors are not a concern for me. My version of the investor update is the dinner chat with my wife. It’s not a fun update most days.

English

1

0

1

18

Dan Loewenherz@dwlz·4d

It may not seem like it, but investor updates are more for the founders than they are for the investors. No one wants to send a “bad” update. The update isn’t the problem though. The problem is your company. The update is just holding a mirror up to your performance. If you don’t like what’s in the mirror, the solution isn’t to stop looking in the mirror, it’s to take steps to look better. If you stop sending updates, you’re tacitly admitting that masking or omitting poor performance makes the poor performance less bad, which is obviously crazy. If I had to guess, I’d bet companies that send frequent updates perform better. Whether this is cause or effect is hard to say, but why take your chances? Just send the update. :)

Philip Johnston@PhilipJohnston

Early-stage founders should send more investor updates. There, I said it. As a (very) small time angel I now see that maybe half of founders send regular updates, which suprised me a bit tbh. At the end of every @YCombinator batch they give a speech which is essentially: “Just don’t die.” Every founder has their own way of interpreting/internalising/actioning that, but mine is: “Religiously send an update at the start of every month.” I never skip and I’m never late. 100% of companies that die, stop sending updates at some point. Its simple to the point of stupidity really: If I’m still sending updates, I’m still alive. Also has some nice externalities like investors are more likely to be there for you when times are tough.

English

3

0

5

792

Dan Loewenherz@dwlz·3d

@shiyam_kashfiq Correct

English

0

10

Shiyam Kashfiq@shiyam_kashfiq·4d

@dwlz Yep, regular updates keep founders honest and moving forward

English

1

0

1

12

Dan Loewenherz@dwlz·4d

@housecor Feeling it too

Dan Loewenherz@dwlz

After people get a few months of seeing the impact of AI-induced slop in their codebases, and the resulting slowdown this incurs, I predict we're going to see a mass reversion back to "handwritten" code. TBH I'm feeling it myself some days. I'm actually faster with Cursor tab and just nailing what needs to happen. AI has a terrible habit of spending time on things that just don't matter. Might be more "work", but in terms of clock time, I'm getting things done more quickly without involving token inference. Maybe I'm weird. Maybe this is a terrible prediction (as things frequently are with things that change so quickly), but in this case I've been observing my behavior and others for months and I'm seeing a slow steady trickle back to the "old ways".

English

0

40

Cory House@housecor·4d

Even in the age of AI, I can do many tasks more quickly manually than via prompting. Entering the right prompt, granting access, waiting for response, reading response, reviewing results, and iterating on results is often more time-consuming than just manually making a change.

English

94

29

373

19.2K

Dan Loewenherz@dwlz·4d

@jorilallo Looks disgusting too, yuck

English

0

1

31

Jori Lallo@jorilallo·4d

Not only is Doordash pushing for their AI generated fake descriptions, now we have shops putting up AI slop photos

English

11

0

35

4.3K

Dan Loewenherz@dwlz·4d

@nikitabier Every candidate’s dream offer

English

0

281

Nikita Bier@nikitabier·4d

Me trying to close a hire

English

939

160

6.3K

690.1K

Dan Loewenherz@dwlz·4d

@ColinGardiner Say more :)

English

0

1

36

Colin Gardiner@ColinGardiner·4d

Florida is a special place

English

2

0

5

286

Dan Loewenherz retweetledi

Christos Tzamos@ChristosTzamos·12 Mar

1/4 LLMs solve research grade math problems but struggle with basic calculations. We bridge this gap by turning them to computers. We built a computer INSIDE a transformer that can run programs for millions of steps in seconds solving even the hardest Sudokus with 100% accuracy

English

239

787

5.9K

1.6M

Dan Loewenherz@dwlz·4d

@maxbranzburg How is this not a $500b company?

English

0

440

Max Branzburg@maxbranzburg·4d

Saving > 3.6%+ APY Spending > 4% BTC back Payments > free, instant, global Earning > stake to earn 7%+ APY Loans > as low as 4% APR Crypto > trade 1M+ assets, perps Stocks > all stocks & ETFs Commodities > gold, silver, oil, etc. Events > sports, politics, etc.

Brian Armstrong@brian_armstrong

Everything you need in one app.

English

55

35

416

210K

Dan Loewenherz@dwlz·4d

Hi San Francisco :)

Català

0

8

237

Dan Loewenherz@dwlz·4d

@PhilipJohnston Exactly :)

English

0

16

Philip Johnston@PhilipJohnston·4d

@dwlz Yep, you wanna be on the right side or that, regardless if it’s correlation or causation!

English

1

0

4

203

Dan Loewenherz@dwlz·5d

@jc_gilbert it doesn't say anything?

English

1

0

1

37

JC Gilbert@jc_gilbert·5d

we need a startup that helps cancelling the subs we don’t even we know we still have i have a $70/year sub on my amex i don’t even know what it is so how am i even supposed to cancel it jesus christ i can’t be the only one with that problem right??

English

3

0

4

214

Dan Loewenherz@dwlz·6d

@scaling01 openai has been doing this since forever.

English

0

255

Lisan al Gaib@scaling01·6d

How in the linear attention did they pull that off?

Lisan al Gaib@scaling01

Anthropic no longer charges extra for longer context windows

English

42

15

889

103.5K

Dan Loewenherz@dwlz·6d

Here are my recommended Radar rules for Stripe that have led to the best tradeoff between preventing fraud and allowing legitimate transactions. These techniques have gotten us to a dispute rate of 0.02% for Star Path, which I'm proud of. This is lower than 70% of businesses in North America, and lower than 90% of similar businesses. (I'm not sure what Stripe considers a "similar business", but I'd be curious to know. Any Stripes who can drop some inside baseball here would be welcome.) Our block rate is also a bit higher than others, but this is by design. We are okay with that. Our goal is to build an amazing product, not financially engineer our way to growth by allowing transactions that we shouldn't let through. If someone really wants to subscribe, but the payment is blocked, they will reach out, or change their payment method. Radar Rules So here are the radar rules we have turned on (any others not listed have so low volume that in practice they do not matter): * Block all payments with a risk score of 75 or higher (I believe this is the default) * Require 3DS on non-US transactions * Block transactions failing CVC verification * Block transactions failing postal code verification * Review transactions with elevated risk level Other Tips First, if you receive an early fraud warning, pay attention to it! You should cancel the subscription (if applicable) and refund the relevant payment immediately. Every time I thought to challenge an early fraud warning, I've looked into the situation in detail and thought to myself: "This user seems legit, I'll let it through". And like clockwork, it's turned into a chargeback. Stripe's systems are good at this! Secondly, review retried or blocked transactions and see if you can notice trends. Are they frequently from certain countries? Do the customers in question have have early-fraud warnings? If so, see if you can make a rule to address it. Having a high frequency of failed transactions is just a distraction and waste of resources; better to block those individuals from even getting to the point of being able to start a trial or subscription. Anyways, back to work 🫡

English

0

2

211

Dan Loewenherz@dwlz·13 Mar

We've been getting a flood of new sign ups and trials for Star Path today. Wasn't sure why. I then got a tipoff from the astrology subreddit, and discovered out that one of the best astronomical data sources (Astro-Seek) just closed off public access due to excessive AI bot abuse. ChatGPT, Claude, Gemini, etc., all rely on this data to "do" astrology, and people are noticing. I expect to see this happen with increased frequency, across all verticals and industries. Once the web starts getting locked down, and more data goes "dark", acquiring such data is going to be a big business.