Every law firm intake form has the same problem: it treats a $50,000 PI lead and a "what is the cheapest will I can buy" tire-kicker the same way. Both go into the same queue. Both wait 4 hours for triage. By the time anyone calls back, the high-value lead has signed with a competitor.
AI qualification fixes this.
The architecture in 4 pieces
1. Intake form. Built natively . 6 questions max — more than 6 and people abandon, less and you cannot qualify. For PI: name, phone, case type, when did the incident happen, were you injured and treated, brief description.
2. Webhook to AI qualifier. When the form is submitted, the platform fires a webhook. The endpoint runs a small AI prompt (Claude Haiku, Gemini Flash, GPT-4o-mini all work, all cost pennies):
Classify this lead as HOT, WARM, COLD, or JUNK.
HOT: clear injury, recent (last 30 days), in-state, treated.
WARM: injury but older incident or unclear treatment.
COLD: no injury, very old, out-of-state.
JUNK: gibberish, spam.
Return JSON: {tier, reasoning_one_sentence, urgency_minutes}
AI responds in under 2 seconds.
3. Routing logic. Based on tier, the intake workflow branches: HOT → SMS attorney personal phone immediately + opportunity in Hot Leads pipeline + booking link to lead. WARM → SMS case manager + booking link. COLD → 60-day nurture sequence. JUNK → tag and discard.
4. The attorney pager SMS. For HOT leads: "NEW HOT LEAD: {{first_name}} | {{case_type}} | {{phone}} | Reasoning: {{ai_reasoning}}. Click to dial: tel:{{phone}}". The attorney can be in court, in a meeting, anywhere — and they are 30 seconds from the prospect.
Why this beats everything else
Manual triage by paralegal: slow (1-4 hrs), inconsistent. Round-robin to attorneys: devalues attorney time, low close rate. Generic auto-responder + 24h follow-up: loses the high-intent window.
AI qualification gives the attorney the high-fit lead at peak intent in 30 seconds. Conversion roughly 2-3× the alternatives.
The cost
Gemini Flash: ~$0.001 per lead. Claude Haiku: ~$0.002. For 100 inbound forms/month, between 10 and 20 cents/month in API costs.
What can go wrong
Two failure modes: AI mis-classifies edge cases (always include a daily-digest fallback for the office manager to review COLD/JUNK); attorney ignores the SMS (the workflow only works if the firm has a culture of dropping everything for HOT lead pings).
Where this fits
One of six workflows in the scale.law Law Firm Intake Stack — alongside missed-call text-back, booking + reminders, prospect nurture, review automation, and referral engine. Delivered white-label to marketing agencies serving law firms, $800 + $400/mo per client. scalelaw.tech/services.