Predict for mid-size plaintiff firms.
Predict gives every case a jurisdiction-tuned settlement estimate with a 90% confidence band, drawn from comparable verdicts and settlements, so every attorney in the firm values a case the same way.
10+ attorneys, 5+ paralegals, intake handled by associates and managed at the partner level. The bottleneck moves from "do we have time to value this case?" to "do four different attorneys value this case the same way?" Predict is what makes the answer yes.
Solo or under 5 attorneys? See Predict for solo and small firms.
Cases — pending review across the firm
| Case | Jurisdiction | Predicted | Band | Status | Owner | Activity |
|---|---|---|---|---|---|---|
| #2026-0419 Henderson v. Acme Transport MVA · rear-end | TX · Harris | $485K | ± $42K | Above threshold | M. Williams | 2h ago |
| #2026-0418 Vargas v. Stratford Plaza PL · slip & fall | FL · Miami-Dade | $72K | ± $34K | Partner review | D. Chen | 4h ago |
| #2026-0417 Liu v. Crestview Apartments PL · premises | CA · Alameda | $148K | ± $24K | Above threshold | M. Williams | yesterday |
| #2026-0416 Okafor v. Beltline Logistics MVA · intersection | GA · Fulton | $58K | ± $11K | Below — refuse | Paralegal | yesterday |
| #2026-0415 Rivera v. Westside Diner PL · slip & fall | TX · Bexar | $94K | ± $48K | Wide band · review | D. Chen | 2d ago |
| #2026-0414 Park v. North Star Freight MVA · rear-end | TX · Harris | $312K | ± $38K | Above threshold | M. Williams | 3d ago |
The mid-size firm's valuation problem is consistency.
An associate who's been with the firm 18 months values a soft-tissue rear-end at $90K. The senior partner who hired them values the same case at $145K. Both are defensible inside their own experience. Neither is auditable. Partner review of every intake is the workaround — and it doesn't scale past the partner's calendar.
Associate variance
Four associates value the same case four different ways. The variance is normal — they're calibrating against different recent dockets — but it's hidden, and it produces inconsistent demand letters across the firm.
Same number, every attorney
One model, one cohort, one band. Associates anchor to the same prediction; partner review focuses on the methodology and the edge cases, not the math. Because every attorney anchors to the same prediction and the same band, the spread between two attorneys valuing the same case collapses to the width of the band, not the width of their experience.
Partner review is the bottleneck
The 11AM partner meeting reviews 40 intakes from the prior week. Each gets 90 seconds. The cases that needed 10 minutes get the same airtime as the ones that needed 30 — and case-selection discipline erodes by Thursday.
Triage before the meeting
Predicted band ≥ threshold and tight → auto-signed by the associate. Wide band or threshold edge → partner reviews. The 11AM meeting shrinks from 40 cases to 8 — the 8 that actually need the senior judgment.
Audit trail lives in email
"Why did we sign the Garcia case?" → six months later, a slack thread, three calendar invites, and a vague memory of a Tuesday meeting. The firm can't learn from its own case-selection decisions because the decisions aren't recorded.
Every prediction logged
Per case: the inputs, the band, the cohort, the attorney who ran it, the decision. Quarterly the firm can look at its own retrospective — which predictions landed inside the band, which didn't, where the firm's threshold should move.
Built to scale with the practice, not punish growth.
Flat $499/mo covers every attorney, paralegal, and intake associate. We don't charge you for growing the team.
Provision the firm in 15 minutes. Deprovision when an associate leaves. Audit logs on by default.
The compliance review is on the security page, ready for the IT diligence the firm runs before a vendor goes live.
Partner review changes shape, not effort.
The senior partner stops re-doing the valuation math on every intake and starts reviewing the methodology: which sub-cohort the model pulled, why the band is tight or wide, whether the associate captured the notice status correctly. The work that needs human judgment gets the human judgment; the work that doesn't, doesn't. The partner time that comes back goes into the cases that pay the firm (depositions, mediations, the trial calendar) instead of intake math.
What we can show you today is the model, not a firm testimonial. On a held-out test set, predictions are 90 to 92% accurate against the realized settlement (median absolute percentage error), trained on 312K verdicts and settlements, each shown with a 90% confidence band. We will publish recovered-partner-hours with its range once we have a full quarter of firm data to measure it against.
Read how this is calculated →How the number compares to what the firm uses today.
Most firms already value cases with a verdict-search database, a senior partner's read, or both. Predict does not replace the judgment. It gives every attorney the same starting number, with its band and the cohort behind it.
What the demo covers.
See Predict run on a case from your docket.
20 minutes, screen-shared, on one of your recent matters. The founder and the methodology lead walk through the prediction, the cohort, and how the firm would deploy. The trial after the demo is 14 days, full feature set, no credit card.