6 ms·

Keenable SELECT: an agent that searches the web in SQL

60 points by phoenix120 17d ago

vivzkestrel 17d ago
400. That’s an error. The server cannot process the request because it is malformed. It should not be retried. That’s all we know. terrible, i wrote a 10 line prompt and I got this error and completely lost my query when I tried to google sign in from a firefox browser on mac
dleeftink 17d ago
Echoes of a semantic Web that never was! Showing the prompt trajectories is also a nice touch.
- matt4711 17d ago
  Hey! I’m on the team working on this. The semantic web comparison is pretty close to how we think about this. One key difference is that we’re using LLMs to create structured data on the fly based on the query. That means we don't need to relying on webmasters to explicitly annotate pages with metadata which in my view was one of the main reasons why the semantic web did not work out.
  - evgenysokov 17d ago
    [dead]
simonw 17d ago
Assuming this is accurate, the diagram showing movement between different AI labs is pretty interesting: https://select.keenable.ai/r/frontier-ai-researcher-moves-since-2025/JTPBEMWoebfxxOGuWyaqQg https://select.keenable.ai/r/frontier-ai-researcher-moves-si... Suggests that when Meta AI started throwing around those multi-million-dollar paychecks they got 12 from OpenAI, 5 from DeepMind, 2 from Thinking Machines Lab and none from Anthropic. (There's a small print disclaimer about accuracy at the bottom: "This report was generated automatically by Keenable SELECT at a user's request, from publicly available web sources linked herein. Keenable does not review, verify, or endorse its contents and makes no representation as to accuracy, completeness, or timeliness; AI-based extraction may contain errors.")
- matt4711 17d ago
  Hey, I'm on the team working on this. In terms of accuracy the best way to gauge the usefulness of these reports is to search for some you are deeply familiar with and then estimate. In general I find these reports have some gaps but for most topics they provide a reasonable picture of what is going on. I personally prefer them over deep research reports coming out of chatgpt etc. as they are just walls of text that I skip to go to the result table. For example as somebody interested in coffee this is pretty comprehensive based on my previous research work: https://select.keenable.ai/r/dual-boiler-prosumer-espresso-machines-price-features-dimensions/w7YkxvKVxXAma9L9vnUYrQ https://select.keenable.ai/r/dual-boiler-prosumer-espresso-m...
logancbrown 17d ago
SQL shall never die, in contrast, it shall grow in scope and utility
NME-64 17d ago
Looks like it thinks rather highly of itself - https://select.keenable.ai/r/the-agent-native-rebuild-of-the-software-stack/O7pKmgCB-ACmTpB389CHEA https://select.keenable.ai/r/the-agent-native-rebuild-of-the.... Otherwise looks really cool!
embedding-shape 17d ago
"One user, inigyou, out-commented all of Hacker News in August 2026 — 3,344 comments, 4.4× the runner-up" - https://select.keenable.ai/artifacts/MarIEMVi6t63M4qovQ2cVw https://select.keenable.ai/artifacts/MarIEMVi6t63M4qovQ2cVw Prompt was "What user on Hacker News made most comments in August, 2026?". Seemed pretty fast, used reasonable sources, results makes sense, pretty good job done by the people building it :) Some others: - "What website/domain was the most popular on the frontpage of Hacker News ..." - https://select.keenable.ai/artifacts/SITl-y08FZxJLnif_c_fSw https://select.keenable.ai/artifacts/SITl-y08FZxJLnif_c_fSw (seems reasonable at a glance) - "How has the sentiment of PG shifted on Hacker News between 2010 and 2026?" - https://select.keenable.ai/artifacts/pmW5bxWt4t1CgniaF8kBeg https://select.keenable.ai/artifacts/pmW5bxWt4t1CgniaF8kBeg (less good, found too little data, moved ahead with an analysis anyways) I wish there was a way to abort/cancel/edit the requests, as I am, like many of you, still human, and therefore do mistakes sometimes, but I do realize them, although sometimes a bit too late.
- ButlerianJihad 17d ago
  inigyou claims (1) to be a sockpuppet of a previously banned user (2) to be now banned/restricted and (3) to have a username based on a slur, which I have not been able to figure out, despite diligent LLM assistance.
- matt4711 17d ago
  Thanks for the feedback on the cancel/abort/edit requests. we are working on this already.
zcw100 17d ago
SEM_EXTRACT, SEM_MATCH, WEB_SEARCH are pretty vague functions. Makes me wonder why you'd bother with the SQL part at that point and not just ask the question.
- matt4711 17d ago
  You need ways to sift through large amounts of unstructured data and do data analysis after. We created these functions to be able to use LLMs to parse unstructured data from the into structured data which we then process with a SQL like engine to map/group/filter etc.
spullara 17d ago
Love seeing people experiment with stuff like YQL (https://en.wikipedia.org/wiki/Yahoo_Query_Language https://en.wikipedia.org/wiki/Yahoo_Query_Language) taking advantage of all the AI stuff definitely increases the power.

60 points by phoenix120 17d ago

400. That’s an error. The server cannot process the request because it is malformed. It should not be retried. That’s all we know. terrible, i wrote a 10 line prompt and I got this error and completely lost my query when I tried to google sign in from a firefox browser on mac

dleeftink 17d ago

Echoes of a semantic Web that never was! Showing the prompt trajectories is also a nice touch.

matt4711 17d ago
Hey! I’m on the team working on this. The semantic web comparison is pretty close to how we think about this. One key difference is that we’re using LLMs to create structured data on the fly based on the query. That means we don't need to relying on webmasters to explicitly annotate pages with metadata which in my view was one of the main reasons why the semantic web did not work out.
- evgenysokov 17d ago
  [dead]

simonw 17d ago

Assuming this is accurate, the diagram showing movement between different AI labs is pretty interesting: https://select.keenable.ai/r/frontier-ai-researcher-moves-since-2025/JTPBEMWoebfxxOGuWyaqQg https://select.keenable.ai/r/frontier-ai-researcher-moves-si... Suggests that when Meta AI started throwing around those multi-million-dollar paychecks they got 12 from OpenAI, 5 from DeepMind, 2 from Thinking Machines Lab and none from Anthropic. (There's a small print disclaimer about accuracy at the bottom: "This report was generated automatically by Keenable SELECT at a user's request, from publicly available web sources linked herein. Keenable does not review, verify, or endorse its contents and makes no representation as to accuracy, completeness, or timeliness; AI-based extraction may contain errors.")

matt4711 17d ago
Hey, I'm on the team working on this. In terms of accuracy the best way to gauge the usefulness of these reports is to search for some you are deeply familiar with and then estimate. In general I find these reports have some gaps but for most topics they provide a reasonable picture of what is going on. I personally prefer them over deep research reports coming out of chatgpt etc. as they are just walls of text that I skip to go to the result table. For example as somebody interested in coffee this is pretty comprehensive based on my previous research work: https://select.keenable.ai/r/dual-boiler-prosumer-espresso-machines-price-features-dimensions/w7YkxvKVxXAma9L9vnUYrQ https://select.keenable.ai/r/dual-boiler-prosumer-espresso-m...

logancbrown 17d ago

SQL shall never die, in contrast, it shall grow in scope and utility

NME-64 17d ago

Looks like it thinks rather highly of itself - https://select.keenable.ai/r/the-agent-native-rebuild-of-the-software-stack/O7pKmgCB-ACmTpB389CHEA https://select.keenable.ai/r/the-agent-native-rebuild-of-the.... Otherwise looks really cool!

embedding-shape 17d ago

"One user, inigyou, out-commented all of Hacker News in August 2026 — 3,344 comments, 4.4× the runner-up" - https://select.keenable.ai/artifacts/MarIEMVi6t63M4qovQ2cVw https://select.keenable.ai/artifacts/MarIEMVi6t63M4qovQ2cVw Prompt was "What user on Hacker News made most comments in August, 2026?". Seemed pretty fast, used reasonable sources, results makes sense, pretty good job done by the people building it :) Some others: - "What website/domain was the most popular on the frontpage of Hacker News ..." - https://select.keenable.ai/artifacts/SITl-y08FZxJLnif_c_fSw https://select.keenable.ai/artifacts/SITl-y08FZxJLnif_c_fSw (seems reasonable at a glance) - "How has the sentiment of PG shifted on Hacker News between 2010 and 2026?" - https://select.keenable.ai/artifacts/pmW5bxWt4t1CgniaF8kBeg https://select.keenable.ai/artifacts/pmW5bxWt4t1CgniaF8kBeg (less good, found too little data, moved ahead with an analysis anyways) I wish there was a way to abort/cancel/edit the requests, as I am, like many of you, still human, and therefore do mistakes sometimes, but I do realize them, although sometimes a bit too late.

ButlerianJihad 17d ago
inigyou claims (1) to be a sockpuppet of a previously banned user (2) to be now banned/restricted and (3) to have a username based on a slur, which I have not been able to figure out, despite diligent LLM assistance.
matt4711 17d ago
Thanks for the feedback on the cancel/abort/edit requests. we are working on this already.

zcw100 17d ago

SEM_EXTRACT, SEM_MATCH, WEB_SEARCH are pretty vague functions. Makes me wonder why you'd bother with the SQL part at that point and not just ask the question.

matt4711 17d ago
You need ways to sift through large amounts of unstructured data and do data analysis after. We created these functions to be able to use LLMs to parse unstructured data from the into structured data which we then process with a SQL like engine to map/group/filter etc.

spullara 17d ago

Love seeing people experiment with stuff like YQL (https://en.wikipedia.org/wiki/Yahoo_Query_Language https://en.wikipedia.org/wiki/Yahoo_Query_Language) taking advantage of all the AI stuff definitely increases the power.