Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
piotrgrudzien
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
11 ms
·
31.
▲
One-prompt ArXiv filter: parse email digest, output three papers
(quickchat.ai)
2 points
by
piotrgrudzien
1y ago
|
0 comments
32.
▲
Did I get a little bit hacked by ChatGPT here?
(quickchat.ai)
4 points
by
piotrgrudzien
1y ago
|
0 comments
33.
▲
Challenges of Building an AI Agent on Top of Shopify MCP
(quickchat.ai)
3 points
by
piotrgrudzien
1y ago
|
1 comments
34.
▲
Connect an AI Agent to Jira Tickets
(quickchat.ai)
1 points
by
piotrgrudzien
1y ago
|
0 comments
35.
▲
Talk to any Shopify store via MCP
(quickchat.ai)
4 points
by
piotrgrudzien
1y ago
|
0 comments
36.
▲
I published this blog post 3 years ago, still no answer to the entropy question
(quickchat.ai)
1 points
by
piotrgrudzien
1y ago
|
1 comments
37.
▲
Connecting Claude Desktop to 3 Shopping AIs via MCP
(quickchat.ai)
1 points
by
piotrgrudzien
1y ago
|
0 comments
38.
▲
We launched our MCP – real-time no-code customizable from our platform
(pypi.org)
3 points
by
piotrgrudzien
1y ago
|
0 comments
39.
▲
The Quickchat AI MCP Server
(github.com)
3 points
by
piotrgrudzien
1y ago
|
0 comments
40.
▲
How Model Context Protocol Works. MCP Explained
(quickchat.ai)
1 points
by
piotrgrudzien
2y ago
|
0 comments
41.
▲
Using MCP to Interact with a Quickchat AI Agent
(youtube.com)
1 points
by
piotrgrudzien
2y ago
|
0 comments
42.
▲
How Model Context Protocol Works. MCP Explained
(quickchat.ai)
3 points
by
piotrgrudzien
2y ago
|
0 comments
43.
▲
Thomson Reuters vs. Ross Intelligence implies the LLM was 'non-generative'
(quickchat.ai)
2 points
by
piotrgrudzien
2y ago
|
0 comments
44.
▲
Generative Search Optimization: How to Rank Higher in ChatGPT?
(quickchat.ai)
2 points
by
piotrgrudzien
2y ago
|
0 comments
45.
▲
by
piotrgrudzien
2y ago
Original post (in Polish): https://www.chess.com/pl/news/view/jan-krzysztof-duda-kryzys... Original interview (in Polish): https://www.youtube.com/watch?v=uXEylRcGl8Y
46.
▲
Jan-Krzysztof Duda on 'Crisis of Faith' and Thoughts of Quitting Chess
(chess.com)
2 points
by
piotrgrudzien
2y ago
|
1 comments
47.
▲
by
piotrgrudzien
2y ago
Or that moats exist but in the old-fashioned way: brand, user experience, great understanding of users, highly optimised sales tactics. In short: no free lunch :)
48.
▲
DeepSeek's Long-Term Effect
2 points
by
piotrgrudzien
2y ago
|
3 comments
49.
▲
O3-mini knowledge cutoff is October, 2023
3 points
by
piotrgrudzien
2y ago
|
0 comments
50.
▲
Lichess is leaving X (formerly Twitter)
(twitter.com)
16 points
by
piotrgrudzien
2y ago
|
4 comments
51.
▲
YC Application Tips
(linkedin.com)
2 points
by
piotrgrudzien
2y ago
|
0 comments
52.
▲
We made Davos AI and 600 attendees used it
(quickchat.ai)
3 points
by
piotrgrudzien
2y ago
|
0 comments
53.
▲
We tested DeepSeek. Here's what you need to know
(quickchat.ai)
4 points
by
piotrgrudzien
2y ago
|
1 comments
54.
▲
by
piotrgrudzien
2y ago
An easy to remember / share link: santababy.ai
55.
▲
Santa Baby AI
(quickchat.ai)
7 points
by
piotrgrudzien
2y ago
|
1 comments
56.
▲
The What, Why, and Wow of AI as a Service
(quickchat.ai)
1 points
by
piotrgrudzien
3y ago
|
0 comments
57.
▲
RAG vs. Fine-Tuning
(quickchat.ai)
2 points
by
piotrgrudzien
3y ago
|
0 comments
58.
▲
Is ChatGPT a real polyglot? Problems with multilingualism of LLMs
(quickchat.ai)
4 points
by
piotrgrudzien
3y ago
|
0 comments
59.
▲
GPTs versus Quickchat AI
(quickchat.ai)
4 points
by
piotrgrudzien
3y ago
|
0 comments
60.
▲
Product Update: AI Guidelines, Human Handoff, New Knowledge Base Sources
(blog.quickchat.ai)
1 points
by
piotrgrudzien
3y ago
|
0 comments
More ›