Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
matchagaucho
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
6 ms
·
31.
▲
by
matchagaucho
9mo ago
He would want us to be stewards of the music... and people still talking about it in 300 years.
32.
▲
by
matchagaucho
9mo ago
The TODO lists are also frequently re-inserted into the context HEAD to keep the LLM aware of past and next steps. And in the event of context compression, the TODO serves as a compact representation of the session.
33.
▲
by
matchagaucho
10mo ago
Seems most likely they used an analog trick, like Varispeed, on the final mix (old Beatles trick from the 60s).
34.
▲
by
matchagaucho
1y ago
Creating Planning Agents seems to force that approach. Rather than define skills and execution agents, letting a meta-Planning agent determine the best path based on objectives.
35.
▲
by
matchagaucho
1y ago
As context window sizes increase and token prices go down, it makes more sense to inject dynamic memories into context (and use RAG/vector stores for knowledge retrieval).
36.
▲
by
matchagaucho
1y ago
It also lacks tool calling :-(
37.
▲
by
matchagaucho
1y ago
A few board members have Google/Salesforce connections. They partner on a lot of tech and markets.
38.
▲
by
matchagaucho
1y ago
/s
39.
▲
by
matchagaucho
1y ago
Every time my wife brings up the skis collecting dust in our garage, I’m reminded of the S.K.I. model — Storage Kills Investment.
40.
▲
by
matchagaucho
1y ago
From what I understood, usage will be largely async, background, AI batch jobs using the latest reasoning models.
41.
▲
by
matchagaucho
1y ago
Also missing in these strict, declarative protocols is a reliance on latent space, and the semantic strengths of LLMs. Is it sufficient to put a agents.json file in the root of the /.well-known web folder and let agents just "figu
42.
▲
by
matchagaucho
1y ago
I've become a fan of https://htmx.org for this reason. A small 10KB lib that augments HTML with the essential good stuff (like dynamic imports of static HTML)
43.
▲
by
matchagaucho
2y ago
The difference, though, is these AI IDE startups are essentially built on a fork of Microsoft VS Code. And MSFT has many end game options to dump free IDEs on the market with integrated AI.
44.
▲
by
matchagaucho
2y ago
If robots.txt is an exclusion file, I can't help but wonder if we just need an agents.json inclusion file in the root to get a "standard" started. A2A looks like a typical enterprise, authenticated, strict DTD style spec. Age
45.
▲
by
matchagaucho
2y ago
Correction: both Login.gov and Id.me will be the IDP options after March 4th changeover. https://www.va.gov/?next=loginModal
46.
▲
by
matchagaucho
2y ago
My main recollection of 18F and the USDS is that they expanded the number of identity logins on VA.gov from one to three during their tenure. During one of these transitions, I lost access to my account and had to go through a full re-valid
47.
▲
by
matchagaucho
2y ago
I've noticed sites like Indeed are still showing job listings scraped from our website 5 years ago. I've not chased down every scraper and submitted a remove request. But I can easily see how 20%+ of job listings are dead-ends.
48.
▲
by
matchagaucho
2y ago
Welcome back! And good timing... our brute force cosine similarity scan on Redis-stored vectors needs some help at the core level (ANN, nearest neighbor, k-mean clustering, anything would be an improvement!)
49.
▲
by
matchagaucho
2y ago
Similarly I've found old-school OCR is needed for more reliability.
50.
▲
by
matchagaucho
2y ago
>2) pushing the frontier further out will likely get more difficult. The upside risk is premised on this point. It'll get so cost prohibitive to build frontier models that only 2-3 players will be left standing (to monetize).
51.
▲
by
matchagaucho
2y ago
There's a growing divide between inference GPUs and training GPUs. Investors first need to ask what that ratio might look like in 10 years. 10-to-1? 100-to-1? Inference-to-Training Assuming for each NVDA training GPU sold there ar
52.
▲
by
matchagaucho
2y ago
It's turtles all the way down. C++ compiles to C which compiles to Assembly... AI will own the next layer... dynamically generating SQL and code to access existing systems.
53.
▲
by
matchagaucho
2y ago
To conflate Slack's T&C faux pas with "self host your own LLM" seems like a stretch. IT Sec and Compliance must read the T&Cs and make better vendor selection choices.
54.
▲
by
matchagaucho
2y ago
Same question... ADS-B transponder surely remains on to prevent in flight collisions / maintain vertical sep. Guessing FAA is just committing to anonymize their public part 135 operators list.
55.
▲
by
matchagaucho
2y ago
The best companies can churn their metrics quarterly. Early days of FB were often driven by MAU. Then P95 performance became a priority, leading to higher DAU.
56.
▲
by
matchagaucho
3y ago
The liability, in this case, is not about innovation, but about cost-reduction and outsourcing to a vendor with (allegedly) no audit and compliance controls.
57.
▲
by
matchagaucho
3y ago
They claim the model was grounded with a 25-shot Chain-of-Thought (CoT) prompt.
58.
▲
by
matchagaucho
3y ago
What will future AI code reviewers think when they see prompts interspersed with tips and threats?
59.
▲
by
matchagaucho
3y ago
QSBS equity ownership is common amongst US bootstrapped businesses; which can change the tax concerns upon sale.
60.
▲
by
matchagaucho
3y ago
Those claims are quite a land grab. If on appeal they focused solely on their "Explore GPTs" store, I wonder if the USPTO would be more lenient?
More ›