So what's the AI part in this? (And I am not being snarky or condescending here, everyone assumes that's the default on internet. I have more or less missed the AI wave and just playing catchup.) Are you indexing LinkedIn API and feeding it to an algorithm? Or are you converting a natural language query to a database query using an LLM?
Show HN: Find AI – Perplexity Meets LinkedIn
61–70 of 78 posts
Re: Show HN: Find AI – Perplexity Meets LinkedIn
#62Reminds me of Buildspace [0].
[0] https://sage.buildspace.so/chat/i-need-help-with-my-idea
Re: Show HN: Find AI – Perplexity Meets LinkedIn
#63Cool but after waiting 2 mins for a one sentence prompt I got; We have analyzed 1681 candidates and found 0 records matching the search criteria. The search was initiated 2 minutes ago and took 1 minute and 42 seconds to complete.
That's not a good experience - sorry! What was the prompt? Feel free to reply it here or email and I can look into it.
We're starting a retro on the thousands of searches people ran today, and will tweak the system based on the results. But, an early takeaway is that some searches failed when people applied a filter that the system doesn't understand.
For example, searching "Find AI startups with 50-100 employees" returns 0 results because Find AI doesn't know headcounts yet. (We'll work on that, though).
Re: Show HN: Find AI – Perplexity Meets LinkedIn
#64Cool but after waiting 2 mins for a one sentence prompt I got; We have analyzed 1681 candidates and found 0 records matching the search criteria. The search was initiated 2 minutes ago and took 1 minute and 42 seconds to complete.
And then a second more simple Query ‘find me people who have posted on Mamba architecture and might be looking for jobs’ and got; We have analyzed 695 candidates and found 0 records matching the search criteria. The search was initiated 1 minute ago and took 50 seconds to complete.
I'd recommend going broader on this one, looking for just people who know Mamba architecture.
Find AI is built right now so that people exist only within a company. So, we don't have profiles or index people as individuals - just as employees. That probably made your search hard, because most employees aren't advertising that they are looking for a job (especially on the data sources we use).
Re: Show HN: Find AI – Perplexity Meets LinkedIn
#65So what's the AI part in this? (And I am not being snarky or condescending here, everyone assumes that's the default on internet. I have more or less missed the AI wave and just playing catchup.) Are you indexing LinkedIn API and feeding it to an algorithm? Or are you converting a natural language query to a database query using an LLM?
Find AI is good at meandering or exploratory searches. "Companies that might sell to ", "People that might become future founders based on ". That's essentially what the AI is doing.
Traditional search engines are built for lookups, and don't handle these more subjective or natural language queries as well.
Re: Show HN: Find AI – Perplexity Meets LinkedIn
#66Thanks everybody for trying this out. I just looked at our logs and we're doing >2k requests per minute to OpenAI right now. Free users only get partial search results. If there are any you want to see run to completion, reply here or email me and I'll mark it to run to completion. (The code PRODUCTHUNT is also available this week for a free month of access).
Congrats on the traffic, I hope you get a high conversion rate :) I assume you do more than one GPT call per search, e.g. rewriting user's query into multiple search queries, summarization of the results, etc.?
It's complex. We make a lot of GPT calls. Lots of them. And, for lots of different purposes.
Re: Show HN: Find AI – Perplexity Meets LinkedIn
#67Great stuff! If a query returns zero results, would it return non-zero results with the paid plan?
No, but on the paid plan you can mark it as "live-updating", and you'd get alerts when new matches happen.
We're doing a postmortem on all the searches with 0 results, but please email me what you are looking for and I can give a better answer.
Re: Show HN: Find AI – Perplexity Meets LinkedIn
#68Confused by these numbers: Software engineers Search completed: less than a minute ago • 1792 candidates analyzed • stopped after 53 matches found We have analyzed 84 candidates and found 31 records matching the search criteria. The search was initiated 1 minute ago and took 30 seconds to complete.
That's a bug. We'll take a look - thanks.
Long story short is that some of this analysis is async. We do a lot of parallelization to make it fast. But, we limit results for free searches to keep costs low. But, sometimes extra matches keep getting found after we try to stop free searches.
Re: Show HN: Find AI – Perplexity Meets LinkedIn
#69Where does the data come from?
Scraping
Re: Show HN: Find AI – Perplexity Meets LinkedIn
#70Earlier quoted context omitted.
We don't scrape LinkedIn. We use some basic tools to infer the LinkedIn profile link for each person's page (e.g. [1]), but we don't actually scrape linkedin. [1] https://usefind.ai/companies/contraption-company/people/phil...
You are using data from that LinkedIn leak from a while ago?
No