Never miss another after-hours or Spanish-speaking caller.
ASL Intake answers every call with a bilingual AI voice assistant that discloses it's an AI, detects the caller's language automatically, runs a structured safety-and-urgency screen, and captures the case facts — then hands your firm a clean, schema-validated case record. And with the consented-SMS path, the caller's answers are encrypted in their browser and stored in your firm's own bucket: the platform never holds the plaintext. Your team stays on the cases that matter while the line answers every call, screens for emergencies, and hands you a clean record — a lead is never lost to a voicemail again.
The problem
Every missed call is a missed case.
Small and mid-size personal-injury and criminal-defense firms lose leads after hours, on weekends, and whenever a Spanish-speaking caller reaches a voicemail box. Generic call centers don't understand a safety screen — and handing a third party your callers' raw answers is a liability of its own.
ASL Intake answers every call, screens for emergencies in the caller's language, and is architecturally unable to read the sensitive answers a caller types into the web form.
The privacy boundary
- Web-intake answers are encrypted in the browser
- Stored in the firm's own S3 bucket, not ours
- We hold only token hashes, receipts, and checksums
- The platform never receives the decryption key
- Per-firm PostgreSQL row-level isolation
Capabilities
An intake team that never sleeps — and never leaks.
Bilingual AI voice
Built on Amazon Nova 2 Sonic with automatic language detection and mid-call English↔Spanish switching.
Safety screen
An immediate-danger question, a 911 redirect, the 988 crisis line, and a custody/warrant/deadline urgency screen — one question at a time, with exact approved Spanish wording.
Mandatory disclosure
A bilingual AI-and-recording disclosure is spoken before any personal facts are collected.
Canonical case record
A schema-validated record with parties and roles, consents (with disclosure version and capture method), transcript segments, urgency, disposition, and practice-area taxonomy.
Firm-owned encryption
Consented-SMS → browser envelope-encryption → your firm's storage; presigned uploads with short TTLs, and a field-level crypto module with blind indexes.
Firm-only mobile app
A native Android admin app ("ASL Connect") with FLAG_SECURE screenshot blocking, no file writes, and a clipboard that auto-clears after 60 seconds.
Practice-area catalog
Localized subcategories across PI (car/truck/motorcycle collision, premises, dog bite, wrongful death) and criminal defense (DUI, drugs, DV, assault…).
API, SDK & webhooks
A REST API with OpenAPI docs (EN + ES), a TypeScript SDK, idempotent ingestion, a transactional outbox, and signed direct-delivery webhooks.
No-risk demo
A non-persistent "Intake Readiness Lab" with fictional scenarios that make zero network, DB, or voice calls — so you can evaluate the flow with no live pipeline.
How it works
Call to case record, safely.
Caller dials in
Twilio routes the call to a private voice gateway; the AI greets them in their language and discloses itself.
Screen & capture
A safety/urgency screen runs first, then the AI captures the case facts through Nova 2 Sonic.
Isolate & validate
A schema-validated canonical record lands in an isolated, per-firm row-secured database.
Deliver, signed
Your firm gets the record via a signed webhook and reviews it in the portal or the mobile app.
Answer every caller. Keep every secret.
Bilingual AI intake that screens for emergencies and hands you a clean case record — without ever reading your callers' raw answers.