Case file — 07416CE1
The idea
“Legal AI for contingency-fee personal injury lawyers — pre-trained on injury case law and settlement databases. Cuts intake from 4 hours to 15 minutes per case. Firms handle 200+ cases simultaneously, each worth $30-35K in fees. We charge $2K/month and save each firm $180K/year in staff costs.”
The panel
Market Assessment EvenUp dominates this space with $1B valuation and the largest personal injury dataset (250k+ verdicts, millions of medical records). They've already solved your core problem—intake automation with 99% accuracy, generating 1,600+ demands weekly for established firms. Your $2K/month pricing directly competes with their entrenched position, but they have network effects from their settlement repository that your pre-trained model can't replicate at launch. Red flag you're ignoring: PI firms are deeply conservative about AI accuracy in legal work. EvenUp's "trained by hundreds of legal and medical experts" credibility took years to build. Your unproven model entering a trust-critical market faces massive adoption friction, especially when firms already trust EvenUp's outputs. Your genuine advantage: The 4-hour-to-15-minute intake claim is compelling if real. If you can prove faster onboarding than EvenUp's current workflow for specific firm types (solo practitioners, rural firms), you have a wedge. But you need proof—not just idea stage.
Your core miscalculation: you're vastly underestimating liability exposure. Personal injury law involves specific damage calculations, precedent sensitivity, and jurisdiction-specific rules that a pre-trained model will get subtly wrong consistently. One bad recommendation on a $100K case creates malpractice exposure that kills your unit economics instantly. You need continuous human validation, which erodes your time-savings claim. The build-vs-buy trap: you'll eventually need custom fine-tuning on each firm's historical cases and local court outcomes. That's not a $2K/month SaaS problem—it's custom services work with terrible margins. You should acquire an existing legal research API (LexisNexis, Westlaw integration) instead of building ML from scratch. No moat here. Any legal tech vendor (Casetext, LawGeex, Thomson Reuters) can bolt this onto their existing platforms in weeks. Your only defensibility is relationships, not technology. What works: focusing on intake classification and conflict-checking is genuinely achievable. That's lower-stakes than settlement prediction and doesn't require you to own liability for outcomes.
Your CAC problem is brutal. You're selling to 5,000 PI firms in the US, most run by 2-3 partners who don't budget for software. You'll burn $500K+ on sales before closing your first deal—these aren't self-serve buyers. Your LTV math assumes 3-year retention, but law firms churn fast when economic conditions shift. Your pricing is anchored to their staff savings ($180K/year), which is backward. They'll negotiate you down to $500/month because the real value—faster case throughput—isn't captured in your model. You're not increasing their revenue per case; you're just freeing capacity they may not monetize. With zero traction and no paying customers, you have maybe 18 months of runway before the math breaks. The unit economics don't matter yet because you haven't validated that firms will actually use this or integrate it into their workflow. Integration friction with their case management systems will kill adoption. What works: contingency-fee firms have genuine cash flow problems and hate admin overhead. If you can prove 15-minute intakes stick and don't create liability issues, you've found a real pain point worth solving.
Timing verdict: Late, but not dead. Personal injury firms have been adopting case management AI since 2023-24. You're entering a crowded field where LexisNexis, Thomson Reuters, and vertical specialists already embed intake automation. Your window closes in 12-18 months as incumbents integrate your exact feature set. Macro pressure: litigation funding contraction. Contingency PI firms are under margin squeeze—insurance defense is getting better at AI-driven settlement prediction, driving payouts down. Firms desperate for intake velocity already bought solutions. You need to justify switching costs against entrenched workflows. Opportunity window: Closing. The arbitrage (staff savings) was real in 2024-25. By 2026, adoption is no longer competitive advantage—it's table stakes. Late movers now compete on price, not innovation. One genuine tailwind: Paralegal shortage is acute. Despite automation, firms can't hire. Your 15-minute intake directly solves an unfillable role gap right now, making replacement easier than hiring. Build fast. Differentiate on vertical specialization (motorcycle injury, construction) or geographic jurisdiction depth, not horizontal speed.
Competitors found during analysis
Live dataEvenUp
$1B valuation raised
Dominant PI AI, settlement repository
Cause of death
EvenUp's data moat is a concrete wall, not a speed bump
EvenUp has 250K+ verdicts, millions of medical records, and generates 1,600+ demand letters weekly. That's not just a head start — it's a self-reinforcing data flywheel where every case processed makes the next one more accurate. Your "pre-trained on injury case law" pitch is vaporware by comparison. You can't pre-train your way to jurisdiction-specific settlement intelligence. You'd need years of real case data flowing through your system to even approach parity, and firms won't give you that data without trusting you first. Classic cold-start death spiral.
Liability exposure eats your margin and your pitch
Your entire value proposition is speed — 4 hours to 15 minutes. But personal injury law involves damage calculations, jurisdiction-specific precedent, and medical record interpretation where being subtly wrong on a $100K case creates malpractice exposure for the firm. The moment you need continuous human validation to cover liability (and you will), your 15-minute claim becomes 45 minutes plus a paralegal review, and your ROI story collapses. EvenUp spent years building credibility with "hundreds of legal and medical experts" validating outputs. You have an idea and enthusiasm.
Your TAM is small and your sales motion is expensive
There are roughly 5,000 PI firms in the US worth targeting, most run by 2-3 partners who are notoriously conservative software buyers. These are not self-serve signups — you're looking at $500K+ in sales burn before meaningful revenue. And your $2K/month pricing is anchored to staff savings, which is the wrong frame. Firms will negotiate you down to $500/month because you're not increasing their revenue per case — you're freeing capacity they may never monetize. At $500/month across a slow-adopting market, your unit economics are underwater before you write your first line of code.
⚠ Blind spot
You're building for the average PI firm, but the average PI firm isn't your customer. The firms that handle 200+ cases simultaneously and would benefit most from intake automation are exactly the firms EvenUp already serves — they're the ones with budget, sophistication, and volume. The firms you could actually win (solo practitioners, small shops doing 30-50 cases) don't generate enough fee volume to justify even $500/month, and they'll never integrate a new tool into their workflow without hand-holding that destroys your margins. Your ideal customer and your accessible customer are two completely different firms, and you haven't noticed.
What would need to be true
Solo and mid-size PI firms (sub-100 cases) must be willing to adopt AI intake tools despite having no existing software stack — meaning you need a zero-integration onboarding path that works outside traditional case management systems.
Your model must achieve jurisdiction-specific accuracy high enough that firms don't require full paralegal review of every output — otherwise your time savings evaporate and you're just an expensive autocomplete.
You must reach $50K MRR within 12 months — before LexisNexis, Thomson Reuters, and Casetext bolt intake automation onto their existing platforms and close your window permanently.
Recommended intervention
Stop trying to be a horizontal PI intake platform. Pick a single high-value injury vertical — construction accidents or trucking collisions — where settlement values are highest ($150K+ average), case complexity creates the most intake pain, and EvenUp's general-purpose model is least differentiated. Build jurisdiction-specific intelligence for the 10 states with the highest volume of that injury type. Partner with a single litigation funding firm that finances those exact cases — they have deal flow, they know which firms need help, and they're motivated to accelerate intake because faster intake means faster returns on their capital. You become the "construction injury AI" that litigation funders recommend to their portfolio firms. That's a wedge EvenUp can't easily counter because they're optimized for breadth, not depth.
Intervention unlocking
5seconds
No account needed. One email, no follow-ups.
Want your idea examined? Free triage or full panel →