A national mass-tort firm runs a Roundup spot during the 6pm news in seven DMAs. Phones light up at 6:04pm. By 9pm the inbound queue has 312 calls. Of those, 248 will not qualify: too late on statute, wrong product (Roundup-branded but not glyphosate-formulation), no diagnosis, or the caller is already represented by another firm. The 64 that do qualify need to be in the case-management system before the lead returns the next day, with field-mapping clean enough to file a short-form complaint.
That workflow is not what most legal-intake content describes. It is volume and filtering, with a deadline.
This is not legal advice. It is a working overview of how mass-tort intake differs from personal injury and medical malpractice, and what AI voice agents have to know to handle it without breaking the case-management pipeline downstream.
Mass-tort intake is screening, not consultation
A med-mal intake might run 25 to 40 minutes because causation has to be probed before the firm spends money on a medical-records review. A car-accident PI call might run 12 to 18 minutes because liability and damages need a basic factual record. Mass-tort intake runs 4 to 9 minutes, and the goal is a clean disposition, not a relationship.
The math forces it. A campaign window, defined as the 30 to 90 days a firm is actively advertising for an MDL, will produce 100 to 500 inbound calls a day. At 18 minutes per call, that is 30 to 150 hours of intake-staff time daily. No firm staffs for that. So the call has to be a structured screen: did you take Drug X between Date A and Date B, did you receive Diagnosis Y, are you currently represented, when did you first learn the drug was implicated.
The "no" rate runs 70 to 85 percent on most active MDLs. The intake needs to deliver "no" within 90 seconds when the answer is obvious (caller never took the drug, statute clearly expired, already represented) and capture the full screen in under 6 minutes when it is not.
The AI has to know which MDLs the firm is in
This is where most generic legal-intake AI breaks. The intake schema for Roundup (MDL 2741, glyphosate, non-Hodgkin lymphoma, Northern District of California) is not the schema for Camp Lejeune (Camp Lejeune Justice Act, federal jurisdiction in EDNC, water-contamination presumptions) is not the schema for hernia mesh (multiple MDLs across Bard, Atrium, Ethicon, each with distinct device-identification requirements).
The AI must know:
- Which active MDLs the firm is participating in
- The master-complaint allegations for each
- The short-form complaint fields (product, dates, diagnosis, treating physician, prior litigation)
- The current cutoff date the firm is enforcing for new cases (often tighter than statute)
- Whether the JPML has issued a transfer order that affects which cases the firm is taking
When the JPML consolidates a new product-liability matter, the firm's intake script changes within days. The AI's prompt and schema have to change with it. A vendor that requires a 2-week sprint to update intake logic is unworkable for mass tort.
Bellwether timing changes which leads are worth capturing
Bellwether trials in MDL set the case-value benchmark for the entire docket. A favorable bellwether verdict pushes settlement values up; an adverse one pushes them down or kills the docket. Firms calibrate their intake aggressiveness to where the MDL is in its bellwether timeline.
Pre-bellwether, firms often want every viable case, betting the docket will settle and they will get fee on volume. Post-favorable-bellwether, firms tighten criteria because they can be selective. Post-adverse-bellwether, firms may stop intake entirely or shift to a discovery-phase only posture.
Intake has to capture whether the caller's facts align with bellwether profiles. If the bellwethers in Roundup were heavy long-term agricultural users with NHL diagnoses within 5 years of last exposure, a casual residential user who used Roundup once 12 years ago is not a bellwether-aligned plaintiff and the AI should flag that for triage. The case might still be filed, but it goes into a different settlement tier.
Lead-plaintiff identification is its own subroutine
For PSLRA securities class actions and some consumer class actions, the lead-plaintiff motion has a 60-day window from publication notice. The court appoints lead plaintiff based on the largest financial interest in the relief sought, combined with adequacy under Rule 23.
If the firm is competing for lead-counsel appointment, intake is hunting for plaintiffs with large losses. A caller with $4 million in losses on Stock Y during the class period is a lead-plaintiff candidate. A caller with $8,000 in losses is a class member, valuable but not strategic.
The AI has to ask the loss-size question early, know the class-period dates exactly, and route large-loss callers to a senior attorney within hours, not days. Generic legal-intake AI does not do this. It treats every caller equivalently and produces a list. Mass-tort intake produces a triaged queue.
For more on how schema differs by practice area, see legal intake schema by practice area.
Serial callers from the same incident are a dedup problem
Camp Lejeune generated an estimated 1.3 million potentially eligible claimants. Firms running Camp Lejeune campaigns receive thousands of calls from veterans, family members, surviving spouses, and adult children who lived on base between 1953 and 1987. A single household can generate four or five calls: the veteran, the spouse, two adult children, a parent.
Intake has to dedup. Not just on phone number (households share lines) and not just on name (common names plus generational suffixes confuse exact-match logic). The dedup keys for mass-tort are typically:
- Service member name + service dates + base assignment
- Date-of-birth + last-four SSN (where collected)
- Address-of-record during exposure window
- Existing-client records pulled from the case-management system at call start
The AI has to query the existing client database before or during the call, not after. A duplicate call that gets logged as a new lead is a problem, because the firm now has two intake records for the same matter, the marketing analytics over-count, and a paralegal has to reconcile manually.
Conflict checks at the intake stage covers the broader workflow, but mass-tort dedup is a sharper version of the same problem.
Class action and MDL mass tort are not the same intake
Federal Rule 23 class actions require commonality, numerosity, adequacy, and typicality. The class is certified or it is not, and individual class members do not have separate cases. MDL mass tort consolidates discovery and pretrial across thousands of individual cases that retain their separate identities and proceed to individual trials or settlements.
The intake is different.
Class-action intake captures enough to confirm the caller is in the class and provides notice of rights. Damages are typically formulaic (statutory damages, pro-rata distribution from a settlement fund). The intake does not need an exhaustive damages workup.
Mass-tort intake captures the full case profile because each case is individual. Damages are case-specific. The treating physician matters. The product-identification matters at the lot or model-number level.
If the AI does not know which kind of matter it is intaking, the intake is wrong. A caller about an opioid case might end up routed to class-action language when the firm is actually pursuing individual MDL cases. The disposition data downstream becomes unusable.
Integration with mass-tort case-management platforms
The mass-tort space has its own platform stack. Generic legal CRMs (Clio, MyCase) do not handle the volume or the campaign-attribution requirements well, so firms use Calleva, OnTrak, Litify, Filevine, or MyCase Mass Tort module.
These platforms have:
- Campaign attribution fields (which TV spot, which DMA, which keyword drove the call)
- MDL designation fields (which active matter the case maps to)
- Master-complaint and short-form-complaint integration
- Settlement-administrator export formats
- Client portal sync for retainer signing and document collection
AI voice intake has to write to these systems with the right field mapping the first time. A vendor that ships data as a generic JSON blob and expects the firm to build an integration is shifting the integration cost to the firm. For mass tort, where intake-platform integration is part of the core stack, that is a non-starter.
CRM integration models for AI voice goes deeper on the integration architecture choices.
Five vendor-evaluation questions for mass-tort intake
If you are evaluating AI voice for mass-tort intake, ask:
- How fast can intake schema be updated when the firm joins a new MDL? Days, not weeks. Mass tort moves quickly when the JPML transfers a new matter.
- Does the system query the case-management database for existing-client records before or during the call? Post-call dedup is too late.
- What is the field-level integration with Calleva, Litify, Filevine, or your firm's specific platform? Ask for an example of the field-mapping document, not a marketing one-pager.
- How does the system handle bellwether-aligned vs non-bellwether-aligned facts? If the answer is "we capture everything and the firm sorts it," that is a "no" disguised as a "yes."
- What is the disposition taxonomy for "no" calls? "Not qualified" is not enough. The firm needs to know whether the call was statute-expired, wrong-product, already-represented, no-diagnosis, or out-of-scope, because those numbers drive marketing-spend decisions.
The contrarian point
The marketing pitch for AI voice in mass tort is usually about handling volume. The actual constraint is not volume. It is schema fidelity at volume.
A firm that runs 300 calls through an AI that produces sloppy field-mapping ends up with 300 records that need paralegal review before they can be filed. The intake throughput has not improved; the bottleneck has just moved to a different desk. The AI looked fast and was actually slow, because the intake-management platform integration was an afterthought.
The firms that get value from AI voice intake in mass tort are the ones that treat the integration with Litify or Filevine or Calleva as the primary feature and the conversation quality as the second feature. Conversation quality matters, but it is not the differentiator. Field-mapping that does not require a paralegal to clean up is the differentiator.
If you want to see how that integration actually works in practice, book a call and we will walk through the schema for whichever MDL your firm is actively in. If you want to test it on a low-volume matter first, start free trial and run a parallel intake against your current process for two weeks. Pricing is on the pricing page. For sibling reading on adjacent practice areas, see medical malpractice intake and family law intake.
Mass tort is not personal injury at scale. It is a different discipline, and intake is where the difference shows up first.