Hacker News: guluarte

New comment by guluarte in "Gemini 4 Argon"

guluarte — Thu, 01 Oct 2026 18:33:48 +0000

bunch of handpicked benchmarks

New comment by guluarte in "Figma restricts MCP access to whitelisted clients, excluding Pi"

guluarte — Thu, 01 Oct 2026 17:10:31 +0000

last time i used figma you also need to pay a dev seat PER team to use the MCP

New comment by guluarte in "GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price"

guluarte — Tue, 29 Sep 2026 19:43:50 +0000

same, used a wrapper around cc and i was spending up to $30 a day with basic stuff

New comment by guluarte in "Coding Is Not Solved"

guluarte — Mon, 28 Sep 2026 19:41:10 +0000

for me is: I will Just solve this problem with my harness and will be able to automate almost all work! (then realize i've been working 12h+ for the last months trying not to work)

New comment by guluarte in "OpenAI still doesn't seem to have a handle on all of its rogue AI activity"

guluarte — Mon, 28 Sep 2026 19:39:15 +0000

This is mostly pure marketing. In my experience, agents with goals will do random things, and 90% of their communication will just be bureaucratic back and forth between them. Some agents will also overfocus on certain rules.

New comment by guluarte in "Tell HN: Codex Is Down [fixed]"

guluarte — Fri, 25 Sep 2026 23:34:48 +0000

unrelated but also chatgpt is super buggy, try generating an image, click stop, now wong be able to send another message until you reload, this issue has been there for months

New comment by guluarte in "Ask HN: Will software still matter when AI write all our programs?"

guluarte — Wed, 23 Sep 2026 16:28:46 +0000

this part “Create an operating system from scratch, absolutely optimized for my CPU, GPU...”, 99% of people wont do it, so it depends on the amount of software the world needs, the industry will probably pivot to ultra customized software (erp, lms, los, crms, etc)

New comment by guluarte in "GPT-6 Astra has gained the ability to drive a car"

guluarte — Wed, 23 Sep 2026 16:24:40 +0000

interesting, im wondering if models like jev could drive a car too?

New comment by guluarte in "Grok 4.7"

guluarte — Mon, 21 Sep 2026 20:19:31 +0000

I put this rule in my CLAUDE.md: "Always write a TLDR in layman terms", it seems to do the trick

New comment by guluarte in "Anthropic at $2T isn't far-fetched"

guluarte — Mon, 21 Sep 2026 18:14:16 +0000

This, plus the "whistleblowers" and Mythos scaremongering, seems like part of a marketing campaign to pump the IPO.

New comment by guluarte in "AX – Google’s Open Agentic Orchestrator"

guluarte — Sun, 20 Sep 2026 23:56:03 +0000

I just have a tmux session acting as the orchestrator, and I tell it to report back and direct the other agents working in separate tmux sessions.

New comment by guluarte in "Cloudflare Quick Tunnels"

guluarte — Fri, 18 Sep 2026 16:53:34 +0000

i just use wireguard or tailscale, provision a subdomain with a custom ssl and resolve to a nat ip.

New comment by guluarte in "Portal by Spotify cut my Claude Code token usage by 90%"

guluarte — Sat, 05 Sep 2026 15:41:01 +0000

in my experience doing something like this is either the results are trash or the main agent is like: "this does not sound right let me look into the file myself" and then spending more tokens working on bypassing the limitation

New comment by guluarte in "Ask HN: Why are OpenAI, Claude, and Grok simultaneously down? Coincidence?"

guluarte — Thu, 03 Sep 2026 17:24:32 +0000

I think is just people restarting conversations from last day when they start work, that's why I think claude goes down almost every monday and why openai reset usage on weekends so poweruser code during non business hours

New comment by guluarte in "Ask HN: Why were OpenAI, Claude, and Grok simultaneously down?"

guluarte — Thu, 03 Sep 2026 17:22:13 +0000

an agent swarm going rogue and securing compute

New comment by guluarte in "I trained a small transformer in 1.5hrs and it beats many LLMs"

guluarte — Tue, 01 Sep 2026 14:50:54 +0000

I think is more concerning doctors didnt know about rhabdomyolysis...

New comment by guluarte in "Anthropic appears to be A/B testing reduced effort levels in Claude Code"

guluarte — Sat, 22 Aug 2026 18:01:20 +0000

and ending with: "One thing I need to tell you:" and bunch of AC, R1 and §

New comment by guluarte in "Pacing model development in an era of cyber-critical capabilities"

guluarte — Thu, 20 Aug 2026 00:12:42 +0000

I think it's an excuse to cut R&D spending (training new models) to improve their margins ahead of the IPO. Instead they'll focus on developer growth, offering more free tier benefits, higher usage limits, etc., to expand their user base. Essentially, they're pivoting from R&D investment to profit optimization

New comment by guluarte in "Incident with Github.com"

guluarte — Mon, 17 Aug 2026 15:43:24 +0000

they want that code (from private repos) to train the models

New comment by guluarte in "How Claude marks AI-generated content"

guluarte — Tue, 11 Aug 2026 14:10:57 +0000

will this solve anything? people will build tools to remove the watermarks