The "truncated job descriptions need formatting work" line jumped out — that's the exact problem on the recruiter side too, just upstream. Most JDs aren't structured enough to score cleanly against in the first place, so matching engines on both sides end up compensating for messy source data. We built JD Intelligence (jdintelligence.dev) to clean that up before it hits any matching layer. Good luck with the recruiter side, that's the harder half.