DLC-AI-2

byHemen Ashodia

--- title: STATEMENT OF WORK --- > AI-Powered Legal Voice Intake Platform > > **Prepared by:** Duncan Lewis Solicitors > > **Date:** June 2026 > > **Version:** 1.0 --- POC Scope > > **Classification:** Commercial in Confidence # Introduction & Background {#introduction-background} Duncan Lewis Solicitors (DLS) is one of the UK\'s leading legal aid law firms, operating across multiple practice areas including Housing, Immigration, Family, Employment, and Community Care. The firm handles high volumes of inbound client enquiries across its Contact Centre and departmental teams. DLS seeks to commission the development of an AI-powered inbound voice agent platform that can handle client calls intelligently, execute legal intake workflows, capture structured data, and escalate to human advisors where required. This Statement of Work defines the full scope, functional requirements, technical architecture, and delivery expectations for a Proof of Concept (POC) engagement, with a clear pathway to full production deployment. The initial POC will focus on the Contact Centre and Housing department (covering Homelessness and Disrepair matter types), with the platform designed from the outset to be extensible across all DLS practice areas through a configurable no-code/low-code portal. # Objectives ## POC Objectives The primary objective of the Proof of Concept is to validate the following: - AI-assisted legal intake workflows operating at production quality - Automated triage of inbound calls with structured outcome capture - Conversational reliability across diverse client demographics and communication styles - Dynamic questionnaire execution driven by configurable logic trees - Scalable operational architecture capable of supporting multi-department expansion - Human escalation orchestration with warm transfer and context handoff ## Strategic Objectives - Establish a reusable AI voice intake infrastructure for the firm - Reduce intake friction and improve client experience from first contact - Enable 24/7 inbound enquiry handling without increasing headcount - Standardise intake data quality and Legal Aid Agency (LAA) compliance - Create a platform that allows non-technical staff to configure new questionnaires and workflows via a self-service portal - Support future expansion into all DLS practice areas with minimal development overhead # Scope of Work ## Phase 1 --- POC Delivery (In Scope) {#phase-1-poc-delivery-in-scope} ### AI Voice Agent --- Inbound Call Handling {#ai-voice-agent-inbound-call-handling} - Inbound telephony integration with the firm\'s existing contact centre infrastructure (SIP/PSTN or cloud telephony provider is British Telecoms with calls routed to users through Cisco UC) - Natural language understanding (NLU) for spoken client enquiries in British English - Accent and dialect handling appropriate for DLS\'s diverse client base - Caller identification workflow --- recognition of existing clients via matter reference, date of birth, or phone number lookup against CRM/case management system - New caller registration and lead capture workflow - Graceful handling of interruptions, silence, background noise, and unclear speech - Professional, concierge-style conversational tone with configurable persona and greeting scripts - Support for call recording and transcription ### Legal Intake Orchestration - Dynamic routing of callers to appropriate intake workflows based on stated matter type - Phase 1 matter types: Housing --- Homelessness, Housing --- Disrepair and Data Claims - Structured intake execution following LAA-aligned question sequences - Real-time eligibility indicators (financial eligibility prompts, merit indicators) based on caller responses - Detection of urgency flags (e.g. imminent eviction, emergency homelessness, disrepair posing immediate health risk) with automatic escalation triggers - Graceful re-prompting for incomplete, ambiguous, or contradictory answers ### Dynamic Questionnaire Engine - Configurable questionnaire execution engine supporting branching logic, conditional routing, and skip logic - Phase 1 questionnaires: Homelessness intake, Disrepair intake and Data Claims intake - Support for single-answer, multi-answer, free-text, date, and numeric response types - Version control for questionnaire configurations - Ability for authorised staff to create, edit, and publish new questionnaires via the end-user portal without developer involvement - Questionnaire templates exportable and shareable across departments ### Human Escalation Workflows - Rule-based escalation triggers (urgency flags, caller distress, legal complexity thresholds, explicit caller request) - Warm transfer to available fee earner or intake coordinator with real-time context handoff - Context summary delivered to receiving agent before transfer completion - Voicemail capture and call back scheduling for out-of-hours calls - Queue management with estimated wait time communication - Fallback handling when no agents are available ### CRM & Case Management System Integration {#crm-case-management-system-integration} - Bi-directional integration with DLS\'s case management system (Indigo CMS) with SQL Backend - Automated creation of new matter records from completed intake sessions - Population of structured intake data into CMS matter fields - Client record lookup and matching to prevent conflict and duplicate matter creation - SQL database integration for lead capture, call logging, and operational reporting data - Webhook or API-based event triggers for downstream workflow automation ### AI-Assisted Intake Summarisation - Post-call AI summary generation covering: matter type, key facts captured, urgency indicators, recommended next steps, and assigned fee earner (if escalated) - Summaries formatted to DLS\'s existing attendance note standards - Summaries automatically attached to the relevant matter record in the CMS - Summary review and edit capability for fee earners prior to formal recording ### Operational Dashboard & Reporting {#operational-dashboard-reporting} - Real-time operational dashboard for Contact Centre supervisors - Metrics to include: call volumes, AI resolution rate, escalation rate, average handle time, questionnaire completion rates, matter type distribution - Individual call logs with playback, transcript, and summary access - Configurable alerts for SLA breaches, high escalation rates, or system errors - Exportable reporting in CSV/Excel and PDF formats - Role-based access control for dashboard visibility ### End-User Configuration Portal - Web-based portal for authorised DLS staff to manage platform configuration - Questionnaire builder: create, edit, version, publish, and retire questionnaires and workflows - Persona and script management: edit greeting scripts, hold messages, and escalation messaging - Routing rule configuration: manage matter type routing, escalation triggers, and out-of-hours settings - User management: role-based access control for portal users - Audit trail of all configuration changes with rollback capability ## Phase 2 POC - Integration with practice areas beyond Housing (Homelessness, Disrepair and Data Claims) --- deferred to Phase 2 - Outbound call capability - Multi-language support (Phase 2 consideration) - Video or live chat channels - Integration with third-party legal research platforms - Full LAA billing submission automation - Mobile application development