piterrro

Recently whenever I see something that's vibecoded and I see nobody taking personal responsibility for it: company entity, linkedin profile links - I leave immediately - it's not worth spending my time if the author was not willing to put their reputation at stake.

I guess this is a universal method. In the time when we can have 100 pages vibecoded in an hour by agents, how do we decide which are actually worth human time?

The agent doesn't feel a notion of cost or time, for them producing slop is free. For us humans, reading slop is at cost of cognitive overload.

I think in 10/20 years from now, access to the Internet will be allowed only upon personal identification. Every website will be allowed to ask about your identity upon serving any content. Thats the only way I see this is going. The internet as it is right now does not have a future if majority of traffic will be done by agents. Thus, the cost of that traffic will have to be put on the users and since displaying ads doesnt make sense to agents, a paid access will be introduced (which is what cloudflare is slowly doing)

For a while now, everytime I see „for AI agents” Im looking for benchmarks with comparable solutions.

The only „for AI agent” solution that will gain adoption is/should be measured by two dimensions: token usage and correctness.

If the solution doesnt use less tokens than generating chart.js code for example - why should I use it?

Same for correctness, if the generated chart spec is correct only 90% of the time - why should I use it?

It’s still early but I think this is the direction we should be thinking about „for AI agents” libraries and projects.

From my experience building rapport only makes sense with your most valuable users in terms of revene. Everything else is noise. Non-paying users are most demanding. Its great to talk to lots of people in mvp or ppmf stage, after that once you nail your icp and start charging, you should make the human support a paid feature.

I’ll tell you even more, in enterprise b2b saas, usually the company paying few thousand per month would have less questions and requests than the one paying few hundreds.

Oh, I meant bits not bytes. You are right, the uuid in psql is 16 bytes, the bigint is 8 bytes.

So I’ll have to buy a license for using my mouse now?

For anyone interested in the topic I suggest reading about snowflake id [https://en.wikipedia.org/wiki/Snowflake_ID] or uuid7 the patterns from the article translate cleanly. The bigint is 64 bytes where uuid is 128. There are other caveats but its all about tradeoffs.

Senior engineers build features without over-specified requirements

To me this already disqualifies the benchmark. That statement is missing the most critical piece about senior engineers: the senior engineers know how to obtain input for their work on their own whether that talking to customers or using metrics. Never ever they come up with stuff on their own - that’s junior behaviour.

Until a coding agent will be able to *gather* the input on its own, its never going to be „senior”

I think „the diagnosis” is over simplification and lots of professionals would disagree that there’s always a single one. As a patient your goal is to eliminate the symptoms of whatever is going on in your system. Often times there could be many reasons for it and only curing one can help you already. The diagnosis is a help tool to choose the roght curation method.

Thus, chasing the „right” diagnosis (whatever that is?) is pointless, as it only the outcome (reducing symptoms, stopping the damage) can tell you if the diagnosis was right, but not the only one right.

It funny to see the community here expects the human body to be treated like a deterministic function: for input X expect output Y - and that transfers to diagnosis - people expect to receive the same diagnosis from different specialists for the same issue.

Given human body complexity, the diagnosis is a compound output of the experience, knowledge gained throughout the career and diagnosis methods/equipment, the title (like Dr) is a certification imposed by the state so its "safe" to let people practice since they passed "the bar" - but that doesn't imply everyone will be treating the same.

Some specialists update their knowledge monthly, some yearly and some don't do it at all, there are so many variables in play here (geo, politics, even weather haha).

Having said that, choosing the specialist is really important, getting opinions about their practice and their speciality, you can only maximize your chance of getting the right diagnosis, but don't expect to get it right just because somebody is called a Dr.

I’ve been using DeepSeek v4 pro for a month now in Kilo Code and its great. Fast, reliable, large context window and cheap as… Did 1,5B tokens this month and cost me 40usd (majority cached, but still).

(I know its a satire, but could be seen as an actual post mortem of the future incident) This report made me realize there's no place for humans, as it is right now, in the process of building software systems in the future. Reading this incident made me dizzy after few paragraphs because of the cognitive context overload and I lost track multiple times.

The problem with these libraries is that you need a strict sdlc including linters, ci, reviews and codong standards and still the language itself is not playing in your favour. The idea is nobel and I like that from time to time somebody tries to solve it but imo its never going to go mainstream if the language itself allows for shortcuts.

Take Go as an example, its syntax is very simple on purpose and there is usually only one way to do things.

Unless we get a new language not much is really going to change in the broader landscape Im afraid.

Anyway kudos for the project, I like to see such innovations

thanks good to know! I felt something was off here

I've setup my own company in Poland in 2015, by submitting a web form, 30 minutes later I could issue a fully VAT compliant invoice. This was not an incorporation however - that was a sole proprietorship, but I could still hire people full time, issue invoices etc.

can someone explain how is this different than feeding the VLM model one page at a time?

I wish there was an HTTP method that directly signals async intent of the server. I know we have 202 Accepted status, but these days there are so many APIs that use the async patterns and each one differs a bit. Having a standard for accepting async jobs with idempotency and notification about results via webhook

Good luck on your new endeavour! Selling to devs is hard, did you consider building in public? That would def help get traction imo. Your point about considering API design and overall architecture would definitely differentiate among the all AI slop out there

Its a generalizatiom but this is a direction Im seeing. I maintain OSS project logdy.dev. Until last year around this time I saw a steady influx of github stars every week (not huge but steady enough to project next few weeks). Since then, the traction pretty much stopped. So my thinking is - in coding agents era people dont need to format their local logs since agents take care of the dev mostly. If I were to start this project again, I wouldnt do it right now since the problem wouldnt be worth solving. The economics moved. This is just my example but I guess similar things are happening out there. So, the question remains, is there a use case for niche oss projects anymore? (By niche I mean something thats not mainstream)

What is the motivation and use case in coding agents era for this project? Dont get me wrong, this question echoes in my head for some time now. Especially with supply chain attacks and open source libraries being entirely written by AI, whats my motivation to use open source libs? Previously it was time-saving, now this incentive vanish since its not me spending more time (but agent)

I feel the same way, reading AI built feature entire output makes me cognitively overloaded as well - I can only do so many throughout the day.

What I found myself doing is operating in two modes: 1. For projects that require my attention, I plan and instruct LLM, when needed will draft some code and ask agent to make it better or finish the mundane part (write code and leave gaps with comments asking agent to finish) 2. Full automode where I use spec driven development and TDD - I only ask for changes based on existing PRD, which agent also have to update. Here I do not look at the code at all.

Seems to be working just fine.

Wait, isnt it implemented because of the sheer number of broswers that could be used at the Zoom’s scale? They could’ve used jsonp too it they wanted to bypass CORS. Using image with different dimensions sounds like the most bulletproof way across multiple devices/OSes/browsers

think of all the poor web devs trying to use multiple threads on top of asynchronous operations. wild.

this is new, thanks for that. I just had a brief talk with Gemini about it and it sparked interest in this model (Durable Objects) which I'm going to dive into deeper.

I honestly should print that comment and hang it on a wall.

…competitively-performant… Care to explain competitively to what?

I didnt say LLMs dont know cf workers, I meant that the cf workers runtime is kind of unique and you cannot push there any code without making it cf workers ready in the first place. If you know what youre doing it should be one time step to connect your hono app to cf workers (so not a huge effort) but still its not like tou can run anything there

I’m running entire leadjobs.dev on cloudflare workers and its kind of unreliable for the traffic it gets - around 100 visitors/day. There are some weird errors in d1 from time to time which i cannot debug since its all black box. Also latencies are greater than I would expect, especially, again for d1. Overall its great value for money to get a globally available, low latency service - but I would think twice before going all in. As a sidenote, I expected that, thus the architecture of the service is build in a way that it abstracts the cf runtime and I can switch to any other infra, be it dedicated or another cloud, in a matter of a day

Lets keep in mind this is cloudflare workers runtime - it only makes sense to deploy small things there, maybe static sites. Unless the agent creates something for cf workers from scratch, asking it to „now deploy to cloudflare” will fail so bad.

This would only work if they would provision docker image deployment, similar to google cloud run, but the still, everything serveless has its own caveats…

This is amazing but for everyone out there wanting to buy and build your own AI rig I recommend connecting to one of mamy inference providers and trying out different models themselves for a while. Costs pennies but can give you a nice preview of what you can get with your own rig. Just a friendly tip.

DeepSeek v4 pro is still 10x cheaper than GLM-5.2 and the quality is still enough for 95% of coding tasks.