AI for mass tort intake: how to qualify high-volume campaign responses

AI for mass tort intake means AI agents that answer every response to a firm’s mass tort advertising within seconds, by phone, text, web chat or lead form, screen each claimant against that campaign’s criteria and chase the documents the claim needs. Qualified claimants go to a person who handles the retainer. The AI collects facts and keeps the queue moving; attorneys and intake staff make every decision about representation.

This guide is for mass tort firms and the intake centers that work for them. If your firm mostly handles single-incident cases, start with our guide to AI intake for personal injury law firms instead.

What does AI for mass tort intake actually do?

Mass tort intake has a different shape from everyday personal injury intake. Work arrives in waves tied to ad spend, so a team sized for an average day falls behind the moment a campaign goes live. Each claimant’s story covers months or years of product use, and the claim depends on records the claimant rarely has in hand.

An AI intake system takes on six jobs:

1. Responds at once. A voice agent answers the tracking numbers on your ads; chat agents reply to forms, texts and social messages.
2. Screens against the right campaign. It identifies which ad the person answered and asks that campaign’s questions.
3. Collects documents. It sends upload links and the firm’s records forms, then follows up.
4. De-duplicates. A claimant who calls, fills in a form and texts back becomes one record, not three.
5. Routes qualified claimants to a person. It books the retainer conversation and hands over a summary.
6. Reports by campaign. Every outcome is tagged to its source.

What it never does: tell someone they have a case, estimate what a claim is worth, give legal advice or sign anyone up.

How do you set mass tort lead qualification criteria for each campaign?

Criteria differ for every litigation. In federal court, related cases pending in different districts can be transferred to one district for coordinated pretrial proceedings under 28 U.S.C. § 1407, the multidistrict litigation (MDL) statute, and the facts that matter are specific to each litigation. So every campaign needs its own criteria, written by the attorneys running it.

The categories stay roughly the same from tort to tort; the specifics change every time, and sometimes partway through a campaign:

• Product or exposure. Which product, brand, form or exposure source, and how the claimant got it.
• Usage dates. When use started and stopped, and how often. People rarely remember exact dates, so the AI asks for anchors: a year, a pregnancy, a move, a job.
• Diagnosis. What a doctor diagnosed, in the claimant’s words, and the treating provider.
• Diagnosis date. It matters for the campaign’s criteria and for limitations questions only an attorney should answer.
• Proof of use. Pharmacy or purchase records, medical records that mention the product, packaging, receipts.
• Prior representation. Whether the claimant has signed with, or is talking to, another firm about this claim.
• State of residence. Where the claimant lives and where use and diagnosis happened, which affects who can take the matter.

Write the criteria as a versioned spec

Treat each campaign’s criteria as a short document with a version number, not a script someone edits on the fly. List the hard gates (answers that end intake, such as a current lawyer), the soft signals (answers that send the file to a person), the qualifying values and the date the version took effect. Every record then shows which version it was screened under, so if criteria widen later, you can find the declines worth a second look.

Ask about health with care

Diagnosis questions touch on the worst news a person has had. The AI should ask plainly, accept “I’m not sure” and never suggest the product caused the condition. Some calls come from a spouse or adult child, sometimes after the person who used the product has died; the AI records who is calling and sends these conversations to a person early.

What does the AI collect at each stage, and what does a person decide?

The clearest way to design mass tort intake automation is to split each stage into facts and decisions. The AI gathers facts the same way at any volume. People make the decisions.

Qualification stage	What the AI collects	What a human decides
First contact	Name, phone, email, the campaign or tracking number, preferred language, and a record that the AI disclosure was given	The approved opening script and disclosure wording
Prior representation	Whether another firm is involved in this claim, and its name if known	How to treat unclear answers and any conflict questions
Residency	Current state, and states where use and diagnosis happened	Which states the firm and its co-counsel accept, and any referrals
Product and usage	Product, form, source, approximate start and stop dates, frequency	Whether vague or partial usage meets the campaign threshold
Diagnosis	Condition in the claimant’s words, diagnosis date, treating doctor and facility	Whether the diagnosis fits the litigation, and any limitations concerns
Proof of use	Which records exist, uploads, and the firm’s records request forms sent for signature	Whether the proof is enough to move forward
Retainer	A booked call with a retainer specialist, reminders and a written summary	The engagement conversation, legal questions and whether to sign the client
After signing	Outstanding records, reminder history, claimant replies	Case evaluation, and what to file and when

Routing qualified claimants to the retainer

When a claimant clears every hard gate, the AI does not try to close. It books the retainer call with a person, live or at the next open slot, and sends reminders. The specialist gets a one-page summary: campaign, criteria version, every answer, documents received and anything flagged.

Route to a person at any stage when the claimant asks for one, asks a legal question (“What’s my case worth?” or “Is there a deadline?”), is calling about someone who has died, is distressed, or gives answers your spec doesn’t cover. A warm transfer with a briefing beats a callback note.

How does mass tort intake automation handle records follow-up?

The intake conversation is often the easy part. Getting proof of use and medical records into the file is slower, and that is where claimants go quiet. An AI agent runs the chase the same way on every file:

1. Right after intake: a text and email listing exactly what is missing, with a secure upload link for that claimant.
2. A checklist, not a pile: each upload is matched to the item it satisfies, so everyone can see what is outstanding.
3. Spaced reminders: a set number, on a schedule your firm approves, each naming the missing item.
4. Help when someone is stuck: the agent explains the firm’s standard steps for getting records, or books a call with staff.
5. Hand off after the last reminder: the file moves to a person’s queue instead of going cold.

Follow-ups go only to people who contacted the firm and agreed to be contacted back, and they stop the moment someone asks. Confirm consent requirements for calls and texts with counsel before you set the cadence.

How do you de-duplicate claimants across channels and campaigns?

Duplicates multiply in mass tort intake. The same person may call after a TV spot, fill in a form from a social ad, respond to two of your campaigns or arrive from more than one lead vendor.

• Match on more than one key. Normalized phone number first, then email, then name plus date of birth. A name match alone is a flag for review, not an automatic merge.
• Merge within a campaign, link across campaigns. Someone answering the same campaign twice is one record. Someone who may qualify for two litigations needs two intake files linked to one contact, each screened on its own criteria.
• Check at the door. Run the match before the first screening question, so a returning claimant picks up where they left off.
• Keep every touch. Store each source on the record, and decide as a firm which one gets credit in reporting.

How should a mass tort intake center QA and report on AI intake?

QA sampling of transcripts

Every AI conversation leaves a transcript or recording, so QA no longer depends on which calls someone had time to hear. A sampling plan:

• Read every conversation for the first few days of a new campaign or criteria version.
• After that, review every decline and escalation, plus a fixed random sample of qualified claimants per campaign each week, sized to what your QA team can actually read.
• Score each against one checklist: AI disclosure given, no legal advice, every required question asked, data matches the conversation, correct criteria version, appropriate tone.
• Feed every miss back into the spec or prompts, and re-test before the change goes live.

Campaign-level reporting

Mass tort marketing is bought by campaign, so report intake the same way. The numbers worth a weekly look:

• Time to first response, by channel and source.
• Completion rate: conversations finished ÷ conversations started.
• Qualified rate: claimants passing every hard gate ÷ completed intakes.
• Cost per qualified claimant: campaign spend ÷ qualified claimants, compared across campaigns, sources and vendors.
• Signed rate by source: signed retainers ÷ qualified claimants, split by ad source, vendor and channel.
• Records completion: signed claimants with complete records, and how long it took.
• Decline reasons: the count for each hard gate. A spike in one reason is a question for whoever runs the ads.
• Duplicate rate by source: how often each source sends someone you already had.

Why build a custom system instead of an off-the-shelf AI receptionist?

Off-the-shelf receptionist apps are usually built for a front desk: one script, one calendar, one inbox. An AI receptionist for mass tort law firms needs more:

• Criteria change per tort, and mid-campaign. You need several versioned question sets running at once, which attorneys can change without waiting on a vendor.
• Campaign attribution. Tracking numbers, form sources and vendor IDs have to reach every record, or the reporting above is impossible.
• Records chasing. Multi-step document follow-up is a system of its own, not a receptionist feature.
• Case management integration. We connect to case management systems and CRMs with an API, such as Litify, Filevine or Salesforce, so each qualified claimant arrives as a complete record with the transcript attached.
• Ownership. On our builds, the client owns the code, prompts and data.

That is agentic AI in practice: agents that book, send, update records and escalate, rather than only answering questions.

What carries over from a real build

Darfoor Law Firm is a Fort Lauderdale, Florida personal injury firm, not a mass tort practice, but its problems are ones mass tort intake faces at higher volume: inquiries on several channels at all hours, manual re-keying and duplicate records. In four weeks we built an AI voice receptionist for the phone and AI chat agents for two Instagram accounts, Facebook Messenger and TikTok, all running one eight-step qualification. From the Darfoor Law Firm case study:

• 4 → 1: Instagram (two accounts), Facebook Messenger, TikTok and phone unified into one intake pipeline.
• 100% of inbound leads captured and answered, regardless of time of day or staff availability.
• 0 duplicate leads entering the pipeline, thanks to de-duplication by phone number and session.
• 5 qualification gates applied automatically before anything reaches an attorney.

Our published pricing starts with a build sprint from $4,500 for one system, typically two to four weeks. If criteria will change with every campaign, our AI partner engagement ($6,500 a month) covers continuous build, monitoring and prompt work.

What compliance guardrails apply to AI for mass tort intake?

This is not legal advice; firms should confirm their state bar’s advertising and solicitation rules, and any guidance on AI, with counsel. The guardrails we build in:

• Inbound only. The system responds only to people who answered the firm’s advertising or contacted the firm. It never cold-calls or cold-texts.
• Disclose the AI. Every conversation opens by telling the person they are talking to an AI assistant.
• No legal advice. Questions about whether someone has a case, what it is worth or whether a deadline has passed go to a person.
• Screen for representation early. If someone already has a lawyer for this claim, intake ends politely.
• Protect health information. Know where medical data is stored, who can see it and how long it is kept.
• Human supervision. Attorneys own the criteria, review declines and approve every script change.

Florida offers a concrete example. Florida Bar Ethics Opinion 24-1, approved on January 19, 2024, allows lawyers to use generative AI with conditions. As summarized by Hinshaw & Culbertson, a lawyer using a chatbot must tell prospective clients they are communicating with an AI program, the chatbot should not give legal advice, and the lawyer is responsible if it gives misleading information or is intrusive or coercive. Other states have their own rules, so check every state your campaigns run in. Florida firms can read more on our page about AI intake for Florida personal injury firms.

Where to start

Pick one live campaign and write its criteria as a one-page spec: hard gates, soft signals, qualifying values and escalation triggers. If it doesn’t fit on a page, neither the AI nor your team will apply it consistently.

Then pull a baseline for that campaign from last month: median time to first contact, responses that never finished intake, and duplicate records in your case management system. Run the AI after hours on that one campaign first, read every transcript in the first week, and add campaigns only once the QA scores hold.