58 law firms bought AI, only one can tie spend to a matter
jkubicki · x · 2026-09-20
Legal industry consultant jkubicki asked 58 law firms and legal departments whether they could tie AI spend to specific matters and measure the return. Everyone had bought AI; only one organization could produce real numbers — even as firms keep paying more for Harvey and Legora, whose valuation race he says has little to do with client outcomes.
He argues the industry skips the work that matters:
- Nobody decomposes workflows by task, role, time, resource, variance and the job to be done — they stop at tasks, so integration stays shallow.
- Everything is efficiency and optimization; nobody hunts net new value.
- Biggest sin: almost nobody writes evals or defines what good looks like for a disclosure schedule or lease abstract, so everything gets graded on vibes and demos never become weekly production systems.
He also slams unqualified "AI advisors" selling doomed pilots, endless "learning your firm" fees, and conflicts-of-interest stacks, notes firms don't know how to interview real AI talent, and hints he may launch a new service line with others.
More from coding & agent
- Open-source semgrep finds code and logs by meaning, not regex, with cross-language matching — udmrzn · 2026-09-20
- Stardock CEO runs a Claude agent to watch his stocks — 3 cents a week — draginol · 2026-09-20
- Building an internet scanner that filters by natural-language concepts, not keywords — Calm_Apple7505 · 2026-09-20
- Blogger delegates apartment hunting to Meta's Muse agent, only shows up for viewings — TianbaoX · 2026-09-20
- Open-source Reddit MCP ditches OAuth setup — just log in via browser and agents can post — Particular_Law_268 · 2026-09-20
- Meta's Alexandr Wang says the Muse hype is "LEGIT" as early users praise the agent — alexandr_wang · 2026-09-20