Can’t disclose too much yet without nuking my competitive advantage, but basically it’s using LLMs to parse lots of scraped data to identify prospective customers in a certain niche - not in the way “these guys might need it” but more like “these guys have explicitly said they need it”, just that the “say” can be in many different channels and formats, and the LLM’s job is to ingest that firehose of trash and pick out the gems.
All in all a pretty basic and boring use-case that I’m sure I’m not the only one doing, but it’s frankly the first use-case I have where the probabilistic and imperfect nature of LLMs is actually fine.
Can’t disclose too much yet without nuking my competitive advantage, but basically it’s using LLMs to parse lots of scraped data to identify prospective customers in a certain niche - not in the way “these guys might need it” but more like “these guys have explicitly said they need it”, just that the “say” can be in many different channels and formats, and the LLM’s job is to ingest that firehose of trash and pick out the gems.
All in all a pretty basic and boring use-case that I’m sure I’m not the only one doing, but it’s frankly the first use-case I have where the probabilistic and imperfect nature of LLMs is actually fine.