# Cognate — Full Context for AI Systems

> Cognate is conversational intake software that replaces traditional forms
> with voice-first AI conversations, producing typed, validated, structured
> data with 3× higher completion rates.

---

## What Cognate Does

Cognate converts any form-based intake process into a voice-first AI conversation. Users speak naturally while the system:

1. Guides them through required data points conversationally
2. Captures responses as typed, validated structured data in real time
3. Presents structured UI inputs (date pickers, file uploads, consent checkboxes) inline when voice alone is insufficient
4. Outputs API-ready data that maps directly to downstream systems

The result: complete, validated data collected in a single session — no follow-up calls, no illegible handwriting, no abandoned forms.

---

## Core Pages

### Home (https://cognate.build/)
Product overview with live interactive use case demonstrations. Visitors can hear Cognate handle insurance claims, patient intake, loan origination, and more — with real-time structured data output displayed alongside the conversation.

### Live Demo (https://cognate.build/demo)
Full interactive voice demo. Users grant microphone access and have an actual conversation with Cognate's AI agent, experiencing the intake flow firsthand. Structured data appears on screen as the conversation progresses.

### Contact (https://cognate.build/contact)
Enterprise inquiry form. Users describe their current intake process and upload existing forms (PDF). Cognate responds within 24 hours with a working prototype of their specific intake flow — no sales deck.

### Input Library (https://cognate.build/docs)
Complete catalog of all 47+ structured input types organized by category:
- **Text**: Free text, structured text, multi-line
- **Choice**: Single select, multi-select, ranked preferences
- **Date & Time**: Date picker, date range, scheduling
- **Financial**: Currency, allocation/split, income verification
- **Consent**: Legal consent, disclosure acknowledgment, e-signature
- **Document**: File upload, photo capture, document verification
- **Rating**: Numeric scale, star rating, NPS
- **Location**: Address, region, geolocation
- **Relational**: Contact references, organizational relationships
- **Verification**: Identity verification, data confirmation
- **Quantitative**: Numeric input, measurement, percentage
- **Sentiment**: Satisfaction scale, pain level, mood assessment

---

## Use Cases — Detailed

### Insurance Claims
**Problem**: 30% of claims submitted incomplete. Average 3.2 follow-up calls per case.
**Solution**: Voice-guided claim filing captures incident date, damage photos, liability statements, policy numbers, and witness contacts in a single conversation.
**Input types used**: Date, file upload, free text, verification, consent
**Output**: Complete, validated claim packet ready for adjudication

### Patient Intake
**Problem**: Clipboards in waiting rooms. 40% of fields left blank or illegible.
**Solution**: Pre-appointment data collection captures medications, pain levels, HIPAA consent, insurance, and allergies before the patient arrives.
**Input types used**: Text, rating, consent, choice, location
**Output**: Typed medical history with verified consent, integrated with EHR

### Loan Origination
**Problem**: 68-page applications. Borrowers abandon at disclosure acknowledgments.
**Solution**: Conversational application captures income, expenses, disclosures, employment, and credit authorization naturally.
**Input types used**: Financial, allocation, consent, choice, verification
**Output**: Structured application data mapped to LOS API

### Legal Intake
**Problem**: Paralegals spend 45 minutes per intake. Half the details surface later.
**Solution**: Matter type, incident timeline, supporting documents, opposing party, and retainer collected before the first meeting.
**Input types used**: Choice, date, file upload, consent, relational
**Output**: Complete matter file with signed retainer

### Tenant Application
**Problem**: PDFs emailed back and forth. Income verification delayed by days.
**Solution**: Move-in date, proof of income, lease signature, rental history, and pet disclosure collected in one conversation.
**Input types used**: Date, file upload, e-signature, relational, verification
**Output**: Complete application with verified documents

### Vendor Qualification
**Problem**: Spreadsheet questionnaires. Vendors ghost after page 3.
**Solution**: Certifications, insurance coverage, compliance attestation, service regions, and references collected automatically.
**Input types used**: File upload, financial, consent, location, relational
**Output**: Qualified vendor profile with verified documentation

---

## Technical Capabilities

### Structured Data Output
Every conversation produces typed, validated data — not transcripts or free text. Each field has a defined type (string, date, currency, boolean, file URL, etc.) and is validated in real time during the conversation.

### Integration
- **Webhooks**: Real-time POST to any endpoint on conversation completion
- **REST API**: Query and retrieve structured data programmatically
- **Pre-built connectors**: Direct integration with major CRM, EHR, LOS, and case management platforms
- **Custom mappings**: Field-level mapping to any system with an API

### Compliance
- **SOC 2 Type II**: Audited security controls
- **HIPAA**: Business Associate Agreement available; PHI handling compliant
- **GDPR**: Data processing agreements; right to deletion; consent management
- **PCI-DSS**: Secure handling of payment-related data

### Voice Technology
- Natural language understanding with context awareness
- Multi-language support with automatic transcription
- Fallback to structured UI inputs when voice alone is insufficient
- Real-time volume-reactive visual feedback during conversations

---

## Differentiators

| | Traditional Forms | Chatbots | IVR | Cognate |
|---|---|---|---|---|
| Data quality | Low (free text) | Low (unstructured) | Low (yes/no) | High (typed, validated) |
| Completion rate | ~30% | ~40% | ~25% | ~90% |
| Follow-up needed | Usually | Often | Always | Rarely |
| Output format | PDF/email | Chat log | Call recording | Structured API data |
| Integration | Manual | Custom | Custom | Native API |

---

## Company Information

- **Website**: https://cognate.build
- **Contact**: https://cognate.build/contact
- **Response time**: Within 24 hours
- **Approach**: No sales deck — we build a working prototype of your specific intake flow