Maybe I'm missing something in this paper, but this seems to me to be just pretty "standard" caching stuff, albeit:
a) very time sensitive
b) huge files
c) scoped per user
Sort of reminds me of video streaming on CDNs for live video (but per user)?
I still think the big win is going to come based on time of use/live capacity. In a pure economics sense you want to charge a lot for inference when it's oversubscribed and far less when it's off peak (see electricity markets).
We have seen this with anthropics peak times, but it's very blunt currently. We also saw this with batch processing back in the day, but that breaks down because agents are 'chatty' and need to send new responses ASAP. You can't wait ages for each response - it would take weeks to do a simple agentic task if you had to wait hours between turn.
So I think what we'll see is async agents queued up, that you can then decide when to run them - either 'immediately' for time sensitive stuff (for more $$$) or 'best effort' where they can be scheduled to run whenever the provider wants to (3am say). If you also have diagnostics that usually agent task xyz takes y tokens total you can do far more efficient scheduling of these. This also reduces the amount of KVcache gymnastics significantly, as you can dedicate that agent task to a certain rack and schedule it all efficiently.
tl;dr I think the issues with inference efficiency need to be solved at a higher abstraction level of per agent "task" not purely on a per chat message basis. If you can schedule a load of agentic use cases off peak you don't need to preempt them because there is spare capacity by nature.
Yes I think you're right. Reinforcement learning is extremely compute heavy, which cursor doesn't have. And X.ai doesn't have the coding agent data anthropic/OpenAI has, but does have the compute.
However, one thing in AI is that while the usage goes up extremely quickly, it tends to go down just as fast. I know a lot of companies that are in the process of switching from Cursor to Claude Code, so in 6-12 months I'm not entirely sure of the data quality/quantity.
Also I think it is telling that they are calling them SpaceX not X. The X brand is absolutely toxic, especially in enterprise.
Not quite. The UK govt has a rolling target of how many EVs are sold as a proportion of all vehicles. The manufacturers get "fined" £15k per ICE car sold over the target. The targets are ramping up rapidly - 22% in 2024, 28% in 2025, 33% in 2026, etc, reaching 80% (AFIAK, keeps changing) in 2030. There's various trading mechanisms in place and what not so it's a bit more complicated than this.
So it's far better to sell EV below cost (Chinese or not) to get more sold than have to a pay £15k for an ICE car.
The Chinese manufacturers are arguably at double advantage here as they have more BEV to sell so it's far easier for them meet the targets, and they can 'sell' the excess to the Western manufacturers (and further subsidise their EVs!).
I'm not personally against it, I got a brand new EV on a lease recently for close to free after all the tax advantages, and it's not like the Western manufacturers didn't have time to prepare...?
The problem is that there is not enough infrastructure for EVs. If you can't charge at home (e.g. you live in a flat), it is hard to live with an EV and it's much more expensive than the ICE.
I had a rental EV while I was there for 6 weeks last summer, it was a pretty low spec jeep model and I stayed at mates places all over England none of which had parking or charging, to tell the truth charging was a bit spotty in town, but if I was just going around the local area the battery was good for a week or more. My take away was I would definitely rent an EV again, but a lot of the older charging infrastructure still sucks, under-provisioned at peak times, and cost 2-3 times what a similar charge would cost here in NZ. I ended up doing most of my charging at the Tesla superchargers on the motorways and at supermarkets in town. I did 2900 miles total and it was about the same cost as petrol in the end, but worth it as the EV was cheaper to rent and was automatic (which renters charge a premium in UK)
Not sure fast charging all the time is good for battery life though. 99% of my driving in NZ is on a normal 10A overnight charge
Local government can quickly change that, if they get their act together. Here in the Hague, there's literally thousands of public chargers available on the city's residential streets. Coupled with the fact that the charging-price is city-mandated at a fixed rate (currently around 35ct/kwh), this gives a perfectly fine solution for most people. (I can charge at home, for 20ct/kwh currently, so that's even nicer)
Not in the UK. Local governments (councils) are going bankrupt and are saddled with an overwhelming burden to pay for adult and child social care. There's no money for much else
What is actually the realistic cost. Covering infra, the charger and the maintenance of everything involved. Power and transfer included, with transfer including any standing charges. And after that you probably want decent margin to well run the business.
I don't expect 10 thousands of the fast chargers in my town.
I'd love to have slow chargers built into the street lights. Not everyone owns a house, and the public charging usually meets or exceeds the petrol price per mile.
If you're referring to DC charging it's going to be pretty expensive. The construction and power electronics for a DC station is going to be in the millions.
For AC the rectifier is inside the car and the L2 chargers is just a fancy plug. Price should just be the base electricity cost.
Governments would do better to try to fix the bureaucracy around installing L2 chargers in shared living spaces. It's a problem they created and it should be on them to fix. But it guess that's harder than impossible mandates and high EV taxes.
If a significant percentage of cars start to become EVs then spaces where people regularly park overnight will get chargers because it will allow whoever is operating them to make a bit of money selling electricity. You don't have to be making a huge profit margin to make it worth your while to have people passively buying ~200kWh/month of electricity from you.
The same applies to workplaces, especially if solar causes electricity to cost less during daylight hours, and then it becomes convenient to get an EV if there is a charger where you park at night or where you park during the day.
that would depend on the infrastucture cost to install such charging and to maintain it? This is kerbside slow charging presumably overnight. Note that spaces in these residental areas are typically not even marked spaces; the worst outcome might be losing more footpath space to charging infra for road users.
Non-fast chargers aren't very big. They can be installed in lampposts, or in lampost-diameter boxes sunk into the pavement (with the socket sticking out at the top)
About 65% of homes are owned (either outright or via mortgage), but I can't estimate how many of them have garage or off street/driveway parking they can charge their car on.
There's usually very little of the garage space available.
Electricity is expensive in the UK (~25p/kWh) But not gas car expensive. It is £1.57/L (£5.94/gallon).
The EV infrastructure is also pretty dang far along, especially compared to the US. Remember that everything in the UK is a lot smaller and closer together than it is in the US. Further, the UK has a functional train system for long distance travel. You can go from the top of Dunnet Head to Lizard Point in a 15 hour drive.
People downvoting me, Look up chargers in plugshare to see just how many there are in the UK, it's a lot. And also correct my math if it's wrong. An 80kWh car costs £20 to fill up. A 55L car, which has about the same range, costs £85 to fill up.
Also if you are able to charge at home you can subscribe to a smart tariff that gives you electricity for 4p/kWh overnight. That’s £3.20 to fill an 80kWh battery that on a modern car will take you up to 320 miles.
It's cheaper to purchase new bike/used car, tax and insure it, service and fuel, than to use train to ride to work.
Annual train ticket form my small town (25 miles from the Zone 1 of London) is over £5,500. Five GRAND. For the pleasure of standing every time and a much higher risk of getting mugged.
£15k will give you REALLY nice bike or pretty new car. After third year you're saving thousands. Of course if you decided to buy something old and used, you're saving from the second quarter of the first year on.
It's only functional because not everyone can afford another car to work.
It can be, depends on a lot of factors. Obviously flying ryanair will often be the cheapest way to go, but if you do any sort of other regular airline trains will quickly start to win out.
And it's not as if you can fly everywhere in england. As soon as you start looking at more oddball flights (for example, london to birmingham) ryanair goes away as an option and all the flights end up super expensive.
Trains, on the other hand, remain cheap for pretty much the entire nation. You can basically go anywhere by train for under £60. A lot cheaper if you book in advance.
That is because of the cost structure of trains vs planes. Trains require a huge amount of infrastructure, and have higher labour costs because they are slower (so the same journey means people work for longer).
Another problem is that fuel taxes are a reasonably equitable means of paying for the roads, and EVs don't have that -- the closest would be vehicle miles * weight or some such.
Equivalents of fuel taxes for EVs have been announced recently - charging directly on a per mile basis.
> The rate of tax will be 3 pence per mile for fully electric cars; this is around half of the 6 pence per mile the average petrol or diesel driver pays in fuel duty.
In my state they decided to tax EVs punitively through our annual registration fee. I already pay an additional $200 for my EV registration, which is $50 more than the average ICE driver pays is gas taxes. I drive considerably less than the average TN driver. Next year it will be $274.
At least in the US, EVs are not heavier than the average driver's vehicle, though they are heavier than other vehicles in their class. And practically all consumer vehicles are nothing compared to semi trucks.
that is absolutely not the problem. We have more than enough li, subject to cost of extraction. New chemistries dont even need it. you need to update your talking points.
This won't happen unless they outright ban non-EV vehicles which is unlikely considering how many people are still using old cars and cannot afford new cars, how many car enthusiasts are there, and not to mention potential lobbying from big oil.
So 20 years from now, all the old gas regular people cars have aged out. You’re left with what, 1 or 2% of enthusiasts cars? Seems like success to me, and fairly inevitable
There will undoubtedly be a death spiral of sorts when it comes to gas stations, refineries, etc. where they become fewer and farther between as less people buy fuel. And that makes it more expensive and inconvenient, so more people buy EVs, which in turn...
Death spiral to gas stations? why? EV cars need to charge somewhere (and on long trip it can’t be at home) and people need to take a break and grab a coffee sometime too. They will change, sure, but certainly not die.
Refineries will be fewer but we do need another products from them also.
Presumably a lot of people will charge at home which significantly cuts down the number of stations needed or the traffic to those stations.
For example, I have 2 gas stations within a mile of my home. They stay pretty busy because people around me constantly need to fill up. I, on the other hand, basically never visit either of those stations since I switched to an EV. I charge at home.
If everyone around me switched to EVs, those stations could not stay in business. There's a grocery store in the same area which makes anything those stations offer obsolete.
Those are the majority of gas stations that die with a mass switch over to EVs. There's a gas station for my hometown without an attached convenience store with 300 people there. There's no way that station stays in service if a significant portion of the community switches to EVs. It already struggles to be profitable as is (I know the owner).
That might be the case in places where most people live in single-family homes with dedicated garage.
Where I live (Spain) that's not the case at all - our towns are very dense. People in big cities tend to live mostly in flats (Europe's highest elevators-per-capita). Even people in the countryside, where it's more common to have a 1-family homes, often don't have a dedicated garage.
Chargers can be anywhere. They are at grocery stores, parking lots, restaurants, I can see the need for a dedicated re fuel station to disappear when charging is ubiquitous.
This is what people don’t get. Charging just means parking. The idea of dedicated charging stations where you stand around doing nothing, maybe buying a candy bar, really only make sense in the context of a fuel which is not literally already everywhere.
I've seen this on the Autobahns: what were just parking spots with unattended bathrooms are becoming little charging stations. Since I don't have an EV yet, I've not stopped at one to see how high-speed the chargers are, but at the very least, I assume that 10-15 minutes would be enough to get you somewhere more efficient/pleasant to wait for a full charge.
Charging stations will only need to be on highways if cities are sensible and build slow charging infrastructure (aka normal wall sockets) in parking spaces. Urban gas stations will be a thing of the past.
A place where you can take a break and grab a coffee is called a cafe, not a gas station.
Also, with Chinese manufacturers increasingly pushing out batteries capable of 1000+ km, you'll be able to charge fully at home for increasingly long road trips.
In 2045 petrol stations will be well on the way the being about as rare as places selling paraffin or special racing car fuel today.
I don't see how this is an interesting bet. No new petrol car will have been sold for 10 years. Places selling fuel for large lorries etc will last a bit longer, but these are already a fraction of the total.
Does anyone have reliable data on the number locations selling avgas in say, the U.S. compared to the number of locations selling automobile grade gasoline?
This has already happened in Norway, where 96% of new cars sold are EVs. They didn’t ban combustion but they did support adoption with subsidies and other incentives
The problem is a 50kWh battery in a car is worth more as a battery than a typical £1500 car.
The lowest end of the market won’t have electric cars unless the batteries are shagged (early Leafs)
And given how insanely cheap petrol is (15p a mile, so £450 for a low mileage runaround) the savings even if electric was free and they weren’t introducing a 3p/mile charge isn’t there.
A £1500 car is £1500 because it's expected you'll need to replace the engine or transmission pretty soon. That can be up to a £4000 job (£2500 on the low end).
And, as it turns out, a brand new 50kWh battery costs around £4000 to manufacture. Used will be cheaper.
With most modern ICE cars everything but the transmission and the engine will fail before those two go out.
Also: I don't think that's the usual case. Plenty of sub 2k cars that will happily keep driving for years (I've had 3 such cars). ~700 mark is where you start seeing 300k mile "finish-them-off"-type cars.
Plenty of EVs will drive for years as well (so long as they have a good thermal system for the battery). So I'm not sure what point is being made.
Saying "It costs a lot of money to replace the battery" doesn't mean much as the battery, even if it has 70% of it's original capacity, is still perfectly functional. Very much the same as the engine which also costs a lot of money to replace.
I just thought the parent comment was unfair to ICE cars to make the EV proposition sound better. I'm a fan of EVs but they are still more expensive to buy.
That said, very cheap ICE cars have a sweet spot where any damage to engine / transmission / clutch+flywheel often means replacing the entire car since the repair cost exceeds the market value of the car.
>A £1500 car is £1500 because it's expected you'll need to replace the engine or transmission pretty soon.
Really not true at all. Care to share your sources for this claim? Anecdotally, I've (plus friend and family) owned plenty of beater cars in that ballpark price, and none had failures needing to replace engine or transmission. Most of their faults came in the electronics (sensors, actuators, fuse box, wire harness) plus suspension, body rust, etc basically the same parts EVs also have.
Meanwhile, if you look up 'EV clinic' postings online, you'll see they find plenty of design failures with European and Korean EVs that are basically ticking timebombs(sometimes literally, hello Stellantis) where electric motors, inverters or battery packs are guarantee to fail in a short timeframe due to various design faults that were entirely preventable. Most common faults with poor EV designs I saw, seem to be the seal of the electric motor stator cooling which fails quickly leading coolant to flood the motor rotor killing it, needing a rebuild.
From what I gather from their analysis', the crux of this issue seems to be that some modern EVs, especially those less premium ones, are cost cut to the extreme in a race to the bottom to maintain shareholder value, both at manufacturing but also at design phase, leading to cut corners everywhere and such issues being a common occurrence and manifesting en-masse after their warranty runs out. From their analysis, IIRC Tesla's powertrains seem to be some of the most reliable and well designed, with the likes of Audi, VW, BMW, Mercedes being less so and Stellantis being trash tier.
Meanwhile, plenty of older ICEs are largely immune form such massive reliability faults, because they benefited from decades of industrial design and development experience done in a past era where race to the bottom cost cutting and planned obsolescence weren't yet a thing. So I wouldn't be surprised when an older 1500$ ICE car will last longer than a 1500$ EV.
Battery prices are still falling though, it's just demand is enormous. But I works fully expect China to start having "compatible replacement packs" being built once the volume is there to support it.
A logical future market is battery-refurbished EVs, just a question of where the crossover point is.
"Gas" prices are hiking up here - its about £1.90 per litre of diesel at the moment and petrol isn't much less.
In contrast, my cheaper 'leccy rate is now about 25% less at 5.2p per kWh than it was. Electricity is weird in the UK - its pinned to the price of gas and is currently (lol) rather expensive "on peak" at 27.87p per kWh and there is a day standing charge of 47.71p. That's from Octopus.
We also have a petrol car - an elderly Renault Clio. It does just run and run and is pretty economic for a pretty shagged out ICE.
My EV is cheaper to run, by far. However, its unlikely the battery will last 20 odd years. I haven't yet sat down and done some whole life costs for ICE vs EV yet.
My Saic MG4 can do 300+ miles on a 100% charge of its 78kWh battery. After two years, it still manages to exceed its WLTP (with care, when required) and I quite like the ridiculous 0-30 acceleration etc.
It sounds like ICE buyers are subsidizing EV buyers. The math starts to get ugly as EVs become a larger percentage of the mix. ICEs will be priced out of the market, but there won't be anyone left to subsidize the EVs. In short, it sounds like buying a car is about to get a lot more expensive in the UK.
In Germany for example(and other EU countries) you get money back from the government on your tax return for your daily commutes to work, if you live far away enough from work to qualify for commuter subsidies. Those subsidies you get no matter which transportation you use, bicycle, train, car, etc. And plenty of people commute by car when their work/home is far away and remote enough for cycle/public transportation to not be very useful or convenient.
Funny how their solution was to subsidize burning fossil fuels for car commutes to the office instead of, oh I don't know, MANDATING WFH!, given Germans are such staunched green environmentalists. Sure, let's turn off nuclear and ban plastic straws, but let's also subsidize the generation of diesel, brake pad and tire fine dust particles we breathe in, for commuting by car to work. We can't forget our car industry lobbyists.
I am guessing its a deduction in calculating your taxable income. You also say it applies to all forms of transport. That is not a subsidy. It is not different from allowing a factory that uses a fuel to deduct the cost of that fuel in calculating their taxable profits.
Some countries literally subsidize gasoline, but even if your country does not it probably supports oil companies with tax breaks ('creating jobs'). There also are lots of untaxed externalities like global warming and exhaust pollution.
They are not so different: it's the same private/public money dynamic but money flowing in different ways. For consumers it doesn't really matter if it's a tax or 'mandatory levy'. For instance: in the EU it's mandatory to pay for disposal of consumer goods, which is priced in with every purchase. That's as close as a 'disposal tax' as you can get.
In the US, at least, we directly subsidize fossil fuels to the tune of billions of dollars a year -- just counting the above-board, direct subsidies through legislation. If you count indirect subsidies, the figures are staggering.
But put a $7500 point-of-sale rebate on an electric car and people go nuts. So the solution is either make the EV subsidy less obvious, like we do for fossil fuels, or try to educate people on where their tax dollars actually go. The former is more feasible, certainly, in the modern political environment where ideology rules over facts.
That is the aim. Its not a bad idea to reduce the number of cars (at least in cities) but the problem is that the British government wants to drastically reduce the number cars without spending on providing more public transport.
As sibling comment posted, if you run a company you can write off the cost of the lease against corporation tax (25% tax saving), VAT (a further 10% saving if you have personal use, otherwise the full 20% if it's purely business), and you don't have to pay income tax on the money (albeit with a small benefit in kind payment).
If you're in the higher tax brackets this means a £200/month lease (say) ends up more like £90/month.
And because of the "new" £3kish subsidy the govt put in, the car finance companies seem to basically apply that as a big discount to the lease value (or it seems that way?). So you could get a brand new ~£30k EV with no upfront payment and a 2-3 year term for <£100/month including maintenance. Mine even came with a free car charger install at home.
This only works for business leases. The employee sacrifices part of their salary in return for being provided a lease car, and both the employee and employer save tax on that money (up to 45% employee, ~15% employer).
For an ICE car the government claws this money back through hefty "Benefit In Kind" taxes placed on employer-provided vehicles, but as an incentive to drive adoption the rates on EVs are only 3% of the car's nominal purchase price (and you only actually pay up to 45% of that 3%).
It doesn't work out "free," but it's typically as cheap to lease a new EV through this scheme as it is to pay the depreciation on a used ICE.
Not sure what schemes are that good (would be interested to see). My company has one, but I ended up buying a used ICE because it worked out to cost about the same as leasing the equivalent EV model of that car. But that might have been the case for the specific car I was looking at (small Volvo SUV).
check out https://www.leaseloco.com/ (there's other sites - I am assuming you are in the UK), they have a lot of deals on. If you want a really good deal then I don't think you want to look for a specific car, you want instead to see what deals are going on and choose between them. It _seems_ that from time to time the manufacturers do a bit of a firesale via leases.
Unfortunately, if your employer uses a "scheme" then the middleman creams off the 15% that the employer would save, then jacks up prices well beyond the market because they have a captive audience.
If the employer leases the car themselves and provides it to you as a benefit, it can be good value -- but then someone has to own the risk of you changing jobs.
I tried getting "close to free" EV on the electric car scheme and considering take home pay cut, all in all, I'm better off buying WITHOUT the scheme, on pcp.
The reasons are:
- the provider is absolutely bonkers, and employing idiots, offering as "in stock" car models twice since updated or discontinued
- the provider is offering the cars at the very steep price, entirely eating the tax advantage
- on top of that id have to pay BIK tax
- and lastly I'd have no choice of choosing the insurance provider, which results in any damage covered by me not covered, and the company itself having absolutely atrocious reviews (ex customers claiming they've been "fined" for the normal wear and tear (eg tiny windscreen chips))
Try using a MoE model (like Gemma 4 26b-a4b or qwen3.6 35b-a3b) and offload the inference to CPU. If you have enough system RAM (32GB is a bit tight tbh depending on other apps) then this works really well. You may be able to offload some layers to GPU as well though I've had issues with this in MoE models and llama.cpp.
You can keep the KV cache on GPU which means it's pretty damn fast and you should be able to hold a reasonable context window size (on your GPU).
I've had really impressive results locally with this.
I'd strongly recommend cloning llama.cpp locally btw (in wsl2) and asking a frontier model in eg Claude code to set it up for you and tweak it. In my experience the apps that sit on top of llama.cpp don't expose all the options and flags and one wrong flag can mean terrible performance (eg context windows not being cached). If you compile it from source with a coding agent it can look up the actual code when things go wrong.
You should be able to get at least 20-40tok/s on that machine on Gemma 4 which is very usable, probabaly faster on qwen3.6 since it's only 3b active params.
Thanks! These things you're mentioning like "You may be able to offload some layers to GPU...", "You can keep the KV cache on GPU..." configured as part of the llama.cpp? I wouldn't know what to prompt with or how to evaluate "correctness" (outside of literally feeding your comment into claude and seeing what happens).
Aside: what is your tooling setup? Which harness you're using (if any), what's running the inference and where, what runs in WSL vs Windows, etc.
I struggle to even ask the right questions about the workflow and environment.
Yes fair enough, but try feeding my comment in :). It should be enough for it to go on. Then ask it to explain the concepts I mentioned and ask it to suggest follow-up questions for you to learn more about llama.cpp/local inference!
I've had best results with opencode. Running locally w/ 64GB RAM and Radeon 9070XT (16GB). NVidia should be easier (CUDA), I'm on Linux full time now but used to use WSL2 all the time and had all this working in it.
In my case, I was also running an ASR model and a TTS model so it was a bit much for my RTX 3090. I opted to offset like 5 layers to the cpu while adding a GPU-only speculative decoding with their 0.8B model.
I suspect this is the best option. Focus on one city pair to start with, _you_ are the courier and find customers on that city pair. Then you can start figuring out how to attract people onto that city pair so you don't have to do it anymore (as there will be demand).
Atlassian just goes from misstep to misstep. I still use their products quite often. The amount of P0 bugs I experience is absolutely crazy:
- Bitbucket workers are hopelessly out of date (self hosted). We've had to put so many random workarounds in especially for Docker, as they don't keep them up to date enough
- I have had a bug in JIRA for years where I can't reorder a new ticket unless I refresh the page
- Every new feature they introduce into JIRA/Bitbucket over the past couple of years just doesn't work.
- I tried their AI stuff on the free trial, didn't work at all, tried to cancel, can't cancel the free trial online and had to write a load of support tickets (of which the support ticket contact form bugged out multiple times).
Anyone have any insight into why things have got so so dysfunctional? Tech debt? Talent leaving? Both? Even 'bad' enterprise software tends to be able to keep the most basic features running, but Atlassian is a whole new category. If you check their 'community' it is just hundreds/thousands of bugs with workarounds.
> I tried their AI stuff on the free trial, didn't work at all, tried to cancel, can't cancel the free trial online and had to write a load of support tickets (of which the support ticket contact form bugged out multiple times).
Absolutely insane that this is legal. The only reason to do this is to trick and abuse customers. It would be trivially easy to legislate away if our government cared to.
Atlassian seems like a typical entrenched big company, albeit an extreme example. They make money by selling to the bosses of their users and being the default name brand for many cases. Once a company gets to a certain size and doesn't directly compete much on quality internal corruption and incompetence can run rampant.
At least for consumer products you can just issue a chargeback if the company does shitty things that prevent you from canceling. I've done that many times. And you can probably file a complaint with the FTC. I've never done that though.
California's law apparently only applied to B2C, but there was an FTC rule that applied to B2B as well which has been paused by a federal appeals court while they consider if the FTC followed the law in making the rule.
I generally agree with this comment, but what option does a decision maker have here? (apart from similar products that probably will end up doing the same things anyway). Are there equivalent scale/functionality products that can truly serve as an option?
I worked there and the answer is the engineering talent isn't great, in addition to being very unfocused, and tons of pointless org churn. Bitbucket pipelines/workers was originally implemented, essentially, by two guys (I know, because I sat 2 rows of desks apart from them!) if that tells you anything. I doubt there was more than one person actively maintaining it for the past decade, if they didn't get laid off recently. That office doesn't even physically exist anymore, and the people are long gone.
A result of their performance-review driven development. Every engineer, product manager and team is focused on one thing only. Get through the reviews. That means building feature after feature, projecting revenue/cost savings that may not materialize.
No one is focused on quality of the feature. It is all about speed. No one stops and thinks that may be users doesn't care about it. Data science teams are focused on cherry-picking data in a way that shows positive impact to show to the leadership that things are working. Engineers and teams are disincentivized to improve existing features, performance issues(unless it impacts revenue/some enterprise customer complains) and sometimes being punished for it. They are also steadily downsizing their support teams to cut costs.
Things are not going to improve. You should move away.
Featureatis. Just keep pumping out features with no thought. Today, probably also AI-coded .
Even in mid-sized projects if you keep pushing for only new features you'll get a similar system. At least my experience in 3 or so midsized projects that I've worked on where nothing else mattered than checking of features from a huge backlog.
Ah, been at a company like that once before. After a while a dedicated team was created to go in and fix broader issues and essentially stop the system from collapsing under its own weight.
The search function in Jira has always been unusable. It’s perhaps the worst part of the entire platform, but nice to see they’re still focused on adding features I will never use.
I've always thought I was the only one experiencing this and felt like I was crazy.
I guess it's "good" to know that I'm not alone.
The amount of times I've searched for a ticket that I know it's there (because I either have it opened in a different tab, or because I just created it), but can't find, it's just way to many.
The results usually seem completely random to me. It's like the feature never made it out of proof of concept territory. The only advantage of all the email noise Jira sends out is that I can usually search my email for what I'm looking for.
I always got sad when I create a ticket and I see the "ticket created" toast, and then I'm like "oh shoot I forgot to add a screenshot" and go to click the toast to go to my ticket but the toast disappeared. Because then I know that I'm gonna waste the next five minutes of my life looking for it.
FWIW Github has similar shitty search interface. Not sure why.
YouTrack's search is one of the main reasons I use it. Nice query language to filter down on any fields, including custom fields, never had an issue finding things. It's great. With the number of useless search functions in so many products, I'm happy that at least my issue tracking does it right.
Jira is buggy as hell these days. Lots of desyncing that forces me to refresh the page. I can have a ticket open on a sprint board and the modal spontaneously closes after a while, forcing me to reopen it frequently. The other week there were tickets that simply refused to show up in their respective sprint board no matter what I did; later the epic magically appeared on the board out of nowhere, then finally the individual tickets themselves reappeared.
Gotta love the value that vibe coding has added to this world.
Atlassian also shutdown their self-hosted offerings. I'm not sure which version they were on with their datacenter edition, when that got cancelled. Part of it might also simply be a lax approach to QA, now that they don't have to support thousands of installations in on-prem environments. When you can just push out an update, your QA has to be much much better.
Jira has had issues like this even before vibe coding was a thing. Over the years, I've had to reload the page so often, I do it almost instinctively now, like some kind of hospitalism.
You can add to this list: Every single input field they have in Confluence and Jira is misbehaving or broken. Apparently, we can't just have a text input widget that works well. Also apparently, this billion dollar enterprise cannot afford to write or use a proper markdown parser, and apparently we, the user lowlives, cannot be trusted with the full "pwer" of basic markdown laugh.
Jira has vim-like bindings for navigating tickets on boards and years later the feature barely works. It has bugs like pressing the j/k keys changes the URL and random fields but stays on the same ticket or doesn't render the newly navigated-to ticket, etc.
> Anyone have any insight into why things have got so so dysfunctional?
My theory is that there is no incentive for things to not be dysfunctional. At my org anyway Atlassian is almost as entrenched as Google, and we can't even conceive of using something different.
I use plane.so for personal projects after learning about it here. It's a lot simpler and the UI takes some getting used to, but it's fast and it works.
It can't be that hard to just dump/export the entire JIRA in one day and migrate it to something else like linear.app? i was already exporting HTML dumps of the entire JIRA and using it in local tool calls to ground agents as far back as last year instead of wrestling with JIRA API to get it to work. This was before linear became popular.
The migration would take 1-2 engineering man-days I suppose. But its money well spent.
There is an export button on Jira. https://youtu.be/-wGRKzYmA7o?t=92 was what I used. For the workspace docs there is also an export button that can export all the documentation for the project(the export would be in HTML). I then used a simple script built with an LLM to convert all of it into markdowns.
Umm? Is there single step Atlassian did it right? It's a cancer of software development the suits force us to swallow while real development and useful documents are outside of their service because it's so stressful to use.
I use it all day every day with Claude Code. I sometimes wonder past code if this has had the biggest impact on my day to day productivity, either having to make do with semi-bad looking reports or have a designer design them (which is slow).
Sort of feel sorry for Figma in a way though, given all the "partnerships" (highlighting their MCPs) and case studies they've done with Anthropic and then they release this. I note there isn't a testimonial from them this time.
I'm surprised how poorly Figma have used "AI" in general - given they were the "gold standard" in taking emerging technologies (WASM etc) and making an incredible product. The Figma Make thing was incredibly underwhelming, I managed to extract the system prompt out and it's basically just Gemini 3 Pro with a design prompt. Perhaps the original team has left?
They are extremely exposed imo. While all the UI/UX designers will continue using it for the forseeable, I strongly suspect a lot of their (A/M)RR was coming from extra seats for PMs, developers, etc to view and export and do commenting on the files - not core designer usage. I think a lot of this just won't happen on Figma as much.
I think you hit the nail on the head - I bet Adobe and Figma get most of their sold seats from people who don’t really need the full tool and are basically just using it as a viewer and to make very small adjustments or notes
> Sort of feel sorry for Figma in a way though, given all the "partnerships" (highlighting their MCPs) and case studies they've done with Anthropic and then they release this. I note there isn't a testimonial from them this time.
Those partnerships and the MCP were intentionally always watered down. It was purely a play for some cheap exposure without providing anything meaningful.
The very obvious thing for the MCP to expose, that everyone asks for, is to be able to create and edit Figma designs. You can't, likely because they're scared it will kill their product. It's one way: Figma->Agent, no Agent->Figma. They will come around to this one day, potentially when it's already too late. Will be interesting to see how long they wait.
>I strongly suspect a lot of their (A/M)RR was coming from extra seats for PMs, developers, etc
their seats system has always been brutal it’s extremely easy to have the seats balloon if you’re not careful and if they’re yearly there is only a 30 day window a year where you can cancel them when the banner to do so appears.
I tried Figma again after a few years expecting that they'd surely have a tool that lets me describe a design and then it generates a Figma design file.
Nope. Figma Make first renders an HTML/React app with your design. Then you could convert to a Figma design file if you have a pro plan. Extremely underwhelming.
There's hardly any difference between using Figma and just designing it with Codex and Claude Code. And now, Claude Design seems to get it right.
I feel like I'm on a different planet when I see this kind of comment.
What if you need to call multiple external APIs at once with complex json? Sure you can call them one after another, but if each take (say) 2s to return (not uncommon IME), then you are in real trouble with only one thread - even if it is just for one "request".
I guess I'm spoilt in .NET with Task.WhenAll which makes it trivial to do this kind of stuff.
> What if you need to call multiple external APIs at once with complex json?
A few years ago, I had a PHP project that had grown by accretion from taking a single complex input and triggering 2-3 external endpoints to eventually making calls to about 15 sequentially. Processing a single submission went from taking 5-10 seconds to over five minutes.
This was readily solved by moving to ReactPHP (https://reactphp.org/), which implements async via event loops. I was able to reduce the 15 sequential external API calls to four sequential loops (which was the minimum number due to path dependencies in the sequence of operations). That reduced the five minutes back to an average of 20-30 seconds for the complete process.
It wasn't using true multithreading, but in a situation where most of the time was just waiting for responses from remote servers, an event loop solution is usually more than sufficient.
Yeah, though AFIAK these event loops still suffer from blocking on (eg) complex json parsing or anything CPU driven (where real multithreading shines).
But regardless I agree, I'm just saying that these kind of patterns _are_ needed in any moderately complex system, and taking the view that "it's great not to even have it" in the core framework is really strange to me. Esp given every machine I have these days has >10 CPU threads and it won't be long before 100+ is normal.
> Yeah, though AFIAK these event loops still suffer from blocking on (eg) complex json parsing or anything CPU driven (where real multithreading shines).
This is only a problem if the JSON parsing is being done inside the event loop itself. The idea here is that you'd have a separate JSON-parser service that the code in the event loop passes the JSON into, then continues executing the other operations in the loop while it awaits the response from the JSON parser.
Just translate anything you'd spawn a parallel thread for into something you'd pass to a separate endpoint -- that's what I was referring to when I said that the poor multithreading can be easily worked around if you're achieving parallelization by orchestration of microservices.
:)
reply