Comment by slopinthebag
1 month ago
You know when someone is singing the praises about AI and they get asked "if you're so much more productive with AI, what have you built with it"? Well I think a bunch of companies are asking this same question to their employees and realising that the productivity gains they are betting on were overhyped.
LLM's can be a very useful tool and will probably lead to measurable productivity increases in the future, at their current state they are not capable of replacing most knowledge workers. Remember, even computers as a whole didn't measurably impact the economy for years after their adoption. The real world is a messy place and hard to predict!
> measurable productivity
Which measure? Like when folk say something is more "efficient" it's more time-efficient to fly but one trades other efficiency. Efficiency, like productivity needs a second word with it to properly communicate.
Whtys more productive? Lines of code (a weak measure). Features shipped? Bugs fixed? Time by company saved? Time for client? Shareholders value (lame).
I don't know the answer but this year (2026) I'm gonna see if LLM is better at tax prep than my 10yr CPA. So that test is my time vs $6k USD.
Time could be very expensive as mistakes on taxes can be fraud resulting in prison time. Mostly they understand people make mistakes - but they need to look like honest mistakes and llm may not. remember you sign your taxes as correct to the best of your knowledge - your CPA is admitting you outsourced understanding to an expert, something they accept. However if you sign alone you are saying you understand it all even if you don't.
These days productivity at a macroeconomic scale is usually cited in something like GDP per hour worked.
Most recent BLS for the last quarter ‘25 was an annualized rate of 5.4%.
The historic annual average is around 2%.
It’s a bit early to draw a conclusion from this. Also it’s not an absolute measure. GDP per hour worked. So, to cut through any proxy factors or intermediating signals you’d really need to know how many hours were worked, which I don’t have to hand.
That said, in general macro sense, assuming hours worked does not decrease, productivity +% and gdp +% are two of the fundamental factors required for real world wage gains.
If you’re looking for signals in either direction on AI’s influence on the economy, these are #s to watch, among others. The Federal Reserve, the the Chair reports after each meeting, is (IMO) one of the most convenient places to get very fresh hard #s combined with cogent analysis and usually some q&a from the business press asking questions that are at least some of the ones I’d want to ask.
If you follow these fairly accessible speeches after meetings, you’ll occasionally see how lots of the things in them end up being thematic in lots of the stories that pop up here weeks or months later.
[dead]
Economy-wide productivity can be measured reasonably well, although there are a few different measures [1]. The big question I guess is whether AI will make a measurable impact there. Historically tech has had less impact than people thought it would, as noted in Robert Solow's classic quip that "You can see the computer age everywhere but in the productivity statistics". [2]
[1] https://www.oecd.org/en/topics/sub-issues/measuring-producti...
[2] https://en.wikipedia.org/wiki/Productivity_paradox
Try agent zero, you can then upload your bank ( or credit card) statements in CSV etc. It then can analyse it
Number of features shipped. Traction metrics. Revenue per product. Ultimately business metrics. For example, tax prep effectiveness would be a proper experiment tied to specific metrics.
[dead]
I used to write bugs in 8 hours. Now I write the same bugs in 4. My Productivity doubled. \s
I hear this every day, and I'm sure its true sometimes, but where is the tsunami of amazing software LLM users are producing? Where are the games that make the old games look like things from a bygone era? Where are the updates to the software that I currently use that greatly increase it capabilities? I have seen none of this.
I get that it takes a long time to make software, but people were making big promises a year ago and I think its time to start expecting some results.
14 replies →
Even better, I write more bugs in 4 hours than I used to in 8.
5 replies →
I bet you the predictions are largely correct but technology doesn't care about funding timelines and egos. It will come in its own time.
It's like trying to make fusion happen only by spending more money. It helps but it doesn't fundamentally solve thr pace of true innovation.
I've been saying for years now that the next AI breakthrough could come from big tech but it also has just a likely chance of comming from a smart kid with a whiteboard.
Well, the predictions are tied to the timelines. If someone predicts that AI will take over writing code sometime in the future I think a lot of people would agree. The pushback comes from suggesting it's current LLMs and that the timeline is months and not decades.
> I've been saying for years now that the next AI breakthrough could come from big tech but it also has just a likely chance of comming from a smart kid with a whiteboard.
It comes from the company best equipped with capital and infra.
If some university invents a new approach, one of the nimble hyperscalers / foundation model companies will gobble it up.
This is why capital is being spent. That is the only thing that matters: positioning to take advantage of the adoption curve.
Yes scaling is always capitol hungry but the innovation itself is not
[dead]
[dead]
I think for a lot of folks it basically comes down to just using AI to make the tasks they have to do easier and to free up time for themselves.
I’d argue the majority use AI this way. The minority “10x” workers who are using it to churn through more tasks are the motivated ones driving real business value being added - but let’s be honest, in a soulless enterprise 9-5 these folks are few and far between.
Sure but why haven’t you seen a drastic increase in single person startups.
Why are there fewer games launched in steam this January than last?
Because very few knows how to use AI. I teach AI courses on the side. I've done auditing supervised fine tuning and RLHF projects for a major provider. From seeing real prompts, many specifically from people who work with agents every day, people do not yet have the faintest clue how to productively prompt AI. A lot of people prompt them in ways that are barely coherent.
Even if models stopped improving today, it'd take years before we see the full effects of people slowly gaining the skills needed to leverage them.
10 replies →
Because ai doesnt work like this “make me money” or “make stardew valley in space”. The hard part is the painful exploration and necessary taste to produce something useful. The number of these kind of people did not increase with ai.
Eg, ai is a big multiplier but that doesnt mean it will translate to “more” in the way people think.
2 replies →
It comes down back to that whole discussion around intelligence becoming cheaper and more accessible but motivation and agency remaining stable.
I’ve worked with a few folks who have been given AI tools (like a designer who never coded in his life, a or video/content creator) who have absolutely taken off with creating web apps and various little tools and process improvements for themselves thanks by just vibecoding what they wanted. The key with both these individuals is high agency, curiosity, and motivation. That was innate, the AI tooling just gave them the external means to realise what they wanted to do with more ease.
These kinds of folks are not the majority, and we’re still early into this technological revolution imo (models are improving on a regular basis).
In summary, we’ve given the masses to “intelligence” but creativity and motivation stay the same.
1 reply →
My guess is that the true impact of this will be difficult to measure for a while. Most "single-person start-ups" will probably not be high-visibility VC-backed, YC affairs, and rather solopreneurs with a handful of niche moonlighted apps each making 3-4 digit monthly revenue.
1 reply →
Haven't you? I have! In another reply, I noted the avalanche of WisprFlow competitors, as just one example.
1 reply →
Huh? Less games launched on steam? First time I hear that. Any source?
But my guess would be: games are closed sourced and need physics. Which AI is bad at.
3 replies →
[dead]
No. They're firing high paid seniors and replacing them with low pay juniors. This is IBM we're talking about.
The "limits of AI" bit is just smokescreen.
Firing seniors:
> Just a week after his comments, however, IBM announced it would cut thousands of workers by the end of the year as it shifts focus to high-growth software and AI areas. A company spokesperson told Fortune at the time that the round of layoffs would impact a relatively low single-digit percentage of the company’s global workforce, and when combined with new hiring, would leave IBM’s U.S. headcount roughly flat.
New workers will use AI:
> While she admitted that many of the responsibilities that previously defined entry-level jobs can now be automated, IBM has since rewritten its roles across sectors to account for AI fluency. For example, software engineers will spend less time on routine coding—and more on interacting with customers, and HR staffers will work more on intervening with chatbots, rather than having to answer every question.
Where does it say those cuts were senior software developers?
Obviously they want new workers to use AI but I don't really see anything to suggest they're so successful with AI that they're firing all their seniors and hiring juniors to be meatbags for LLMs.
This just doesn't make any sense. Juniors + AI just does not equal seniors, except for prototyping greenfield projects. Who knows about 2 months from now, it moves fast and stuff, but not right now.
2 replies →
probably aren't going to find a lot of articles discussing how water is wet, either.
1 reply →
Meh, i think a lot of companies just wanted an excuse to do lay-offs without the bad press, and AI was convinent.
“AI will steal your job” never made sense. If your company is doing bad, sure maybe you fire people after automating their job. But we’re in a growth oriented economic system. If the company is doing good, and AI increases productivity, you actually will hire more people because every person is that much more of a return on investment
> "if you're so much more productive with AI, what have you built with it"
If my boss asked me a question like this my reply would be "exactly what you told me to build, check jira".
If you want to know if I'm more productive - look at the metrics. Isn't that what you pay Atlassian for? Maybe you could ask their AI...
As a senior engineer sometimes the system shows I did nothing because I was helping others. sometimes I get the really hard problem -'the isn't speller teh' type bugs are more common than thread race conditions - but a lot faster to solve.
[dead]
No one has built business AI that is flat correct to the standards of a high redundancy human organization.
Individuals make mistakes in air traffic control towers, but as a cumulative outcome it's a scandal if airplanes collide midair. Even in contested airspace.
The current infrastructure never gets there. There is no improvement path from MCP to air traffic control.
It's hard work and patience and math.
[flagged]
Everytime someone say something like that there is no link to the product. Maybe because it doesn't exist ?
Historically in a lot of niches such as search marketing etc, people would not name their successful projects because the barrier to entry is low.
It someone can use AI to make a $50,000/year project in three months, then someone else can also do so.
Obviously some people hype and lie. But also obviously some people DID succeed at SEO/Affiliate marketing/dropshipping etc. AI resembled those areas in that the entry barrier is low.
To get actual reports you often need to look to open source. Simon Willison details how he used it extensively and he has real projects. And here Mitchell Hashimoto, creator of Ghostty, details how he uses it: https://mitchellh.com/writing/my-ai-adoption-journey
Update: OP posted their own project however. Looks nice!
2 replies →
What an awful comment. The person above you is now flagged because of your paranoia. Of course later they post a link to exactly what they built.
1 reply →
[flagged]
1 reply →
[flagged]
He is overwhelmed with customers. Can't risk any more awareness.
1 reply →
Sounds nice, for how many years have you had that annual recurring revenue so far?
I only started charging customers in September. Super-linear growth. I launched annual subscriptions and within less than a week > 15% of customers switched.
I'm with you. I own a business and have created multiple tools for myself that collectively save me hours every month. What were boring, tedious tasks now just get done. I understand that the large-scale economic data are much less clear about productivity benefits, in my individual case they could not be more apparent.
I'm thirding this sentiment!
I run an eComm business and have built multiple software tools that each save the business $1000+ per month, in measurable wage savings/reductions in misfires.
What used to take a month or so can now be spat out in less than a week, and the tools are absolutely fit for purpose.
It's arguably more than that, since I used to have to spread that month of work over 3-6 months (working part time while also doing daily tasks at the warehouse), but now can just take a week WFH and come back with a notable productivity gain.
I will say, to give credit to the anti-AI-hype crowd, that I make sure to roll the critical parts of the software by hand (things like the actual calculations that tell us what price an item at, for example). I did try to vibecode too much once and it backfired.
But things like UIs, task managers for web apps, simple API calls to print a courier label, all done with vibes.
1 reply →
Has anyone noticed Amazon or AWS shipping features faster than their pre-GenAI baseline? I haven't
I'm noticeably faster shipping.
The only thing the comments told me is that people lake judgement and taste to do it themselves. It's not hard, identify a problem that's niche enough for a problem you can solve.
Stop arguing on HN and get to building.
Every hype AI post is like this. “I’m making $$$ with these tools and you’re ngmi” I completely understand the joys of a few good months but this is the same as the people working two fang jobs at the start of Covid. Illusionary and not sustainable.
[flagged]
I built and debugged an embedded stub loader for Rp2350 to program MRAM and validate hardware status for a satellite. About 2.5 hours of my time, a lot of it while supervising students/doing other things.
This would have been a couple day+ unpleasant task before; possibly more. I had been putting it off because scouring datasheets and register maps and startup behavior is not fun.
It didn’t know how to troubleshoot the startup successfully itself, though. I had to advise it on a debugging strategy with sentinel values to bisect. But then once explained it fixed the defects and succeeded.
LLMs struggle in large codebases and the benefit is much smaller now. But that capability is growing fast, and not everything software developers do is large.
I'm not doubting of you or anything, but you just proved point above by saying you have a successful project without even mentioning which project is that.
[flagged]
Cool! Can we see it?
[flagged]
Nice, yeah I feel like there's a big opportunity for tech workers who are product-adjacent to use LLMs to get up to speed building SaaS etc.
Are you worried by any of those claims about SaaS being dead because of AI? lol
[flagged]
4 replies →
Details would help your argument. Since many did the same thing, before the AI wave...
Is the business 3 months old now?
It's not an argument, it's a fact.
4 replies →