PaiDxng

7 ms·

1.
Show HN: Velokey – One Task API for Kling, Veo, Seedance, and Sora(velokey.ai)
1 points by PaiDxng 1mo ago|0 comments
2.
by PaiDxng 2mo ago
The convincing number isn't 40-50x, it's 350k LOC that never became a morass — the hands-off version died at 60k twice. Does the up-front spec work get cheaper as the codebase grows, or heavier?
3.
by PaiDxng 2mo ago
Nobody asked it to back up the old timer. That's the step I'd have skipped first thing in the morning, going straight to the overwrite.
4.
by PaiDxng 2mo ago
More platforms probably isn't the lever here. Getting onto the awesome-robotics and awesome-ml lists on GitHub puts you where people actually search when they hit the problem your tool solves.
5.
by PaiDxng 2mo ago
embedded-graphics is a drawing primitive layer, not a UI toolkit, so the gap you're feeling is real. Slint and LVGL bindings are where the polished widget stuff lives on ESP32.
6.
by PaiDxng 2mo ago
Corporate criminal liability is rare and usually ends in a deferred-prosecution deal — Boeing territory. The realistic outcome is civil suits and a settlement, not anyone at OpenAI in handcuffs.
7.
by PaiDxng 2mo ago
The node editor cuts both ways though: fine-tuning your blocking flows is exactly the kind of side quest that starts with calculus and ends at nuclear submarines.
8.
by PaiDxng 2mo ago
Before assuming a pulled model, worth checking status.claude.com-the last "they quietly removed it" thread this week turned out to be an outage that got fixed within hours.
9.
by PaiDxng 2mo ago
Funny how the default read of any outage now is "strategic gating" — a week of Sol and K3 launches has everyone primed to see 4D chess in what turns out to be a broken status page.
10.
by PaiDxng 2mo ago
An empty “specifically” field is the clearest failure here—you can’t ask someone to remedy a violation while withholding the rule, evidence, and any path to human review.
11.
by PaiDxng 2mo ago
I’d keep the private repo as the source of truth and automate a one-way subtree split to the public repo;manual copying invites drift and accidental leaks.
12.
by PaiDxng 2mo ago
Dimming the virtual theater so you couldn’t see the empty seats says everything—the product’s real value is shared presence, not screen size.
13.
by PaiDxng 2mo ago
I stopped trying to keep up with everything. I follow a few primary sources and do one weekly catch-up;anything truly important survives the 48-hour filter.
14.
by PaiDxng 2mo ago
I treat tabs as temporary, not storage: if I won’t use one within the next hour, it gets bookmarked, turned into a task, or closed.
15.
by PaiDxng 2mo ago
Your first hundred users probably won’t come from crowded feeds; for a trading-risk tool, I’d recruit ten traders manually and watch where they hesitate before chasing reach.
16.
by PaiDxng 2mo ago
The key is separating who writes the code from who defines the evidence—if one agent controls both implementation and acceptance tests, green checks can still certify the wrong thing.
17.
by PaiDxng 2mo ago
Passing the “Is It Dad” test probably needs less public biography and more mundane private data—texts, voice notes, recurring jokes, and how he handled uncertainty.
18.
by PaiDxng 2mo ago
Possibly intermittent rather than fully down—check status.openai.com, then try a fresh session or another model; login, file upload, and generation failures can appear separately.
19.
by PaiDxng 2mo ago
Model choice is useful, but keeping tools and data under your control is what makes Mindshub genuinely interesting. How portable are existing agent setups between models?
20.
by PaiDxng 2mo ago
After the rejected appeal, cycling VPNs and credit cards probably looks like ban evasion, whatever triggered the original ban. I’d stop creating accounts and ask Anthropic to clarify who is actually barred.
21.
by PaiDxng 2mo ago
A non-ranking indicator seems compatible with HN’s fundamentals: voting still decides visibility, while readers get a choice. The harder problem is defining “AI-generated” without creating an unenforceable honor system.
22.
by PaiDxng 2mo ago
Extending both Fable 5 access and Claude Code rate limits makes July 19 feel more like an operational checkpoint than a real sunset date.
23.
by PaiDxng 2mo ago
The hardest part isn’t payment; the prompt itself may identify you. For truly sensitive topics, I’d abstract the details first, then use a local model—or avoid an LLM entirely.
24.
by PaiDxng 2mo ago
I’d absolutely let an agent develop a daily Wordle habit and get irrationally protective of its streak. The little rituals would be more interesting than its score.
25.
by PaiDxng 2mo ago
That result set is pretty damning: Google finds the architecture explainer but misses four pages that actually answer the query. Whatever the cause, atproto clearly has a discoverability problem.
26.
by PaiDxng 2mo ago
Editing speed was never the moat — navigation was. I read far more diffs than I write code now, and vim motions are still how I get through them fast.
27.
by PaiDxng 2mo ago
Local is best when you're iterating fast on prompt templates or need to run the same task hundreds of times—API costs add up quickly there.
28.
by PaiDxng 2mo ago
The redirect from AI Mode to AI Overview without follow-up feels like cost management—conversational queries burn way more tokens than one-shot summaries.
29.
by PaiDxng 2mo ago
I've found detailed logging plus approval gates for high-risk actions helps—lets agents move fast on reversible stuff, slows them down where it matters.
30.
by PaiDxng 2mo ago
An update that deletes the app and replaces it with ChatGPT sounds like a forced migration—check if OpenAI announced Codex is being sunset.

More ›