Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
willahmad
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
7 ms
·
31.
▲
by
willahmad
1y ago
according to some sources TikTok US generates $10B. 14B for 10B revenue is a steal. Snapchat generates less revenue (~6-7B), with stalled growth, costs 14B in public markets
32.
▲
TikTok to Be Sold for $14B
(bloomberg.com)
6 points
by
willahmad
1y ago
|
9 comments
33.
▲
by
willahmad
1y ago
They might have dozens of reasons, but they already did what they did. Some of the reasons could be: - mitigation of US AI supremacy - Commodify AI use to push forward innovation and sell platforms to run them, e.g. if iPhone wins local int
34.
▲
by
willahmad
1y ago
China is winning the hearts of developers in this race so far. At least, they won mine already.
35.
▲
Ask HN: How do you orchestrate multiple agents?
1 points
by
willahmad
1y ago
|
0 comments
36.
▲
China's DeepSeek says its hit AI model cost just $294,000 to train
(reuters.com)
5 points
by
willahmad
1y ago
|
2 comments
37.
▲
by
willahmad
1y ago
This can happen because of training data. Imagine you have thousands of legal documents rejecting things to Iran. eventually, model generalizes it and rejects other topics
38.
▲
by
willahmad
1y ago
It can happen because training data contains lots of rejections to groups (Iran sanctioned, don't do business with Iran and so on). Then model might be generalizing 'rejection' to other types of responses
39.
▲
by
willahmad
1y ago
"I will read it after this 15 min short video"
40.
▲
by
willahmad
1y ago
you are right, it probably won't fix the traffic, but it doesn't matter because if you are not steering the wheel then you can work or study while in the traffic. there is a huge economic impact
41.
▲
by
willahmad
1y ago
it's not only about number of users. You also have regulations in different cities and states, existing models trained based on past supply demand behavior, somewhat optimized workflows for the ride hailing industry, payments, risks. T
42.
▲
by
willahmad
1y ago
moat is in institutional knowledge of operations in ride hailing industry. Also, pleasing the customer, imagine opening Waymo app and not being able to order a taxi 40% of the time. With Lyft/Uber you can easily switch the ride mode an
43.
▲
by
willahmad
1y ago
From market standpoint I am curious if Google or Amazon is willing to acquire Lyft for the tech and customer base. Lyft is a good distribution channel for their self driving car initiatives with good coverage, Uber is too expensive at this
44.
▲
by
willahmad
1y ago
Here's my experience with these tools: Good: I can prototype things very quickly thanks to these tools Bad: After couple of vibe coding iterations, I don't have a mental model of the project. Good: When I open my past projects whe
45.
▲
by
willahmad
1y ago
I think what they are trying to do is a type of forward deployed engineer, but with no employment. More they train such engineers more profitable for them to spread the word. In 1st cohort, they're probably going to accept extrovert pe
46.
▲
by
willahmad
1y ago
It could have been, but you still need to compress and transform OpenAPI, here are some examples: * Stripe OpenAPI has ~300 endpoints, imagine adding it to 128k context window, you will run out of context * Most APIs return unix timestamp,
47.
▲
by
willahmad
1y ago
I wish we get better alternatives to Anthropic sooner. Fortunately, OSS models like GLM, Qwen are catching up Obviously, it's good for them if things are regulated, but bad for all of us.
48.
▲
by
willahmad
1y ago
First of all, congratulations Honest question expecting your honest response. How much of it's written by AI? Recently, I noticed I am liking thoughts of real people over the super structured, emdash-y, zero grammar mistake AI texts. I
49.
▲
by
willahmad
1y ago
But they provide sources/links, which I click occasionally and they have utm source with chatgpt
50.
▲
by
willahmad
1y ago
Depends on the questions. If a technical problem, before my questions were leading either to official doc or stackoverflow, now I get possible solutions to try Other types of questions like reviews, product comparisons GPT shows couple of r
51.
▲
by
willahmad
1y ago
From a personal experience, at least 50% of my Google searches moved to ChatGPT and Claude. Since, I saw many similar transitions with my friends and coworkers, I was expecting higher numbers for OpenAI. Do you see a similar transition in y
52.
▲
by
willahmad
1y ago
I have built similar gateway but as a desktop application which runs locally and gives you a lot of control over enabled MCP tools and allows you to have a separate profiles for different combinations of MCP servers and credentials (e.g. pr
53.
▲
by
willahmad
1y ago
Trust takes time to build and only a moment to lose. That’s why we're trying to build the trust by committing to transparency and security. Right now, we’re undergoing the CASA Tier 2 security assessment for our desktop application. Th
54.
▲
Show HN: Aiko – Easiest way to manage MCP configurations and credentials (Beta)
(getaiko.app)
4 points
by
willahmad
1y ago
|
2 comments
55.
▲
by
willahmad
1y ago
I have built bunch of MCP servers so far, and here is my take (and why I love it) There are 2 types of actions, (1) read-only and (2) MCP servers/tools with side effects 1. Read-only types. As you know, LLMs are static models, they don
56.
▲
by
willahmad
1y ago
I am working on easiest way to manage all your MCP configurations and credentials. Additionally, I have added couple of interesting MCP servers, like converting any OpenAPI v3 spec to MCP server (including support of authentication): https
57.
▲
Ask HN: I am looking for MCP server beta testers
2 points
by
willahmad
1y ago
|
0 comments
58.
▲
by
willahmad
1y ago
Would you prefer a local only tool? As a privacy conscious person, I am developing a desktop/local only app and collecting feedback from beta testers. I would appreciate if you can participate as a beta tester and share your feedback.
59.
▲
Qwen3 235B beats Claude on some code benchmarks
(huggingface.co)
21 points
by
willahmad
1y ago
|
2 comments
60.
▲
by
willahmad
1y ago
Thanks for taking time to explain this. > But, I suspect, if the model is able to handle language at all, you'll always be able to get a representation of the prompt out in a text form if I understand correctly, system prompt "
More ›