ippadmin/hermes-recovery
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
master
hermes-recovery/references/dre-specs.md
T

7.8 KiB
Raw Permalink Blame History

DRE Portal — Technical Specifications

Internal working document — July 2026

Claim Numbering

  • Format: DRE-YYYY-NNNN (e.g. DRE-2026-0001)
  • Auto-generated on claim submission

Client ID

  • Format: CLT-YYYY-NNNN (e.g. CLT-2026-0001)
  • Generated on account creation

Closed Case Binders

On case close, two PDF binders are auto-generated:

Binder 1 — DRE Internal: DRE-YYYY-NNNN_CLOSED_DRE.pdf Includes: intake, evidence, AI analysis, AI debtor research, AI weakness analysis, internal notes, signed/notarized LPOA, certified mail receipts, communication log, settlement, fee disclosure, payment confirmation

Binder 2 — Client Package: DRE-YYYY-NNNN_CLOSED_Client.pdf Includes: claim summary, signed/notarized LPOA, settlement/payment confirmation, fee statement. EXCLUDES: AI analysis, debtor research, internal notes.

Evidence Archive: DRE-YYYY-NNNN_evidence.zip — original uploaded documents

Full Workflow

Phase 1 — Intake

  • Client submits claim at debtrecoveryexperts.com
  • Intake form: debtor info, amount, description
  • Document upload (contracts, invoices, correspondence)
  • Terms of Service signed via DocuSeal
  • Claim enters DRE queue

Phase 2 — AI Review

  • AI analysis is per claim, not per client. Each claim gets its own independent:
    • Claim analysis score (0-100)
    • Debtor research / skip tracing
    • Weakness assessment
  • If a repeat client submits a new claim against a different debtor, it gets fresh analysis independent of previous claims
  • AI Re-analysis on new document upload is scoped to that specific claim only
  • Team reviews → Approve / Request More Docs / Reject

Phase 3 — Legal Setup

  • LPOA sent to client via DocuSeal
  • Client signs → notarized via Proof (online RON)
  • DRE now authorized to collect

Phase 4 — Tiered Recovery

  • Tier 1: Soft Touch — email + ACH link (Day 1-5)
  • Tier 2: Formal Demand — certified mail via LetterStream (Day 7-14)
  • Tier 2.5: Lien Threat — pre-lien notice for construction claims (Day 15-21)
  • Tier 3: Escalation — final notice (Day 21-30)
  • Tier 4: Legal Action — referral to partner law firm (Day 30+)

Phase 5 — Settlement

  • Payment collected via Stripe ACH
  • DRE fee deducted per tier schedule
  • Costs deducted (notary, certified mail, filing fees)
  • Balance disbursed to client
  • Case closed → two binders generated

Fee Structure (proposed)

  • Tier 1: 20-25%
  • Tier 2: 30%
  • Tier 2.5: 30% (+ attorney if lien filed)
  • Tier 3: 33%
  • Tier 4: 10% DRE referral + 25% law firm

Search & Document Retrieval

Cases are searchable in the portal via database fields:

  • Claim number, client name, debtor name, amount, status, date range
  • Tags (e.g. "construction", "lien filed", "litigated", "debtor paid Tier 2")
  • Key flags from AI analysis

On case close, PDF text is extracted and stored in the database for full-text search. Future: Paperless-ngx if volume exceeds 500+ cases.

Approval Workflow

  • Approvers: Germaine + Tony — both must sign off before case moves forward
  • Anita: Read-only access — can view pending approvals but cannot approve
  • Reminders: Every 60 minutes, pending approval notification sent to both approvers
  • Approval status visible in claim header: "Pending Germaine · Pending Tony · Acknowledged ✓" Any edit to a claim submission is tracked with a full history:
  • What field was changed (old value → new value)
  • Who made the change (user name or "system" for automated updates)
  • When the change was made (timestamp)
  • Why the change was made (optional note field)

The audit log is visible in the DRE internal dashboard under each claim:

DRE-2026-0042 — Change History
Jul 8, 2026 10:15 AM  |  John (DRE)  |  Amount: $12,000 → $12,450  |  "Corrected from invoice"
Jul 7, 2026 02:30 PM  |  System      |  Status: Submitted → AI Review  |  Auto
Jul 7, 2026 02:28 PM  |  Robert Smith |  Initial submission               |

Edits are allowed on open claims. Closed claims are read-only. The audit trail is included in the DRE Internal Binder on case close.

Repeat Clients

Clients create an account on first claim. On return:

  • ToS already on file (signed once)
  • Client info pre-filled on new claims
  • Dashboard shows: active cases, past cases with outcomes, payment history
  • Can download their own case binders
  • Optional loyalty pricing after 3+ claims (e.g. 25% vs 30%)

Client Portal View (after login)

  • Active Cases: claim number, debtor, amount, current tier
  • Past Cases: claim number, debtor, amount, outcome (Recovered / Uncollectible)
  • [Submit New Claim] button
  • [Download Case Binders] for closed cases
  • [Payment History] — all disbursements received

Required Services

  • DocuSeal: ✅ deployed at sign.itpropartner.com
  • Proof.com: ❌ need account (online notarization)
  • LetterStream: ❌ need account (certified mail)
  • Stripe Connect: ❌ need keys (ACH payments)
  • Law firm partner: ❌ need referral agreement (litigation)

Email Configuration

  • email@redacted — Client-facing. Display name: "Debt Recovery Experts". Used for DocuSeal signing requests, claim status updates, fee receipts, portal login links.
  • email@redacted — Debtor-facing. Display name: "Debt Recovery Experts". Used for demand letters, ACH payment links, collection correspondence.
  • email@redacted — General business inquiries.
  • All mailboxes hosted on MXroute (heracles.mxrouting.net). Outgoing email relayed through mail.germainebrown.com:2525 (MXroute SMTP blocks our IP).

AI Chatbot — Knowledge Base

The DRE chatbot runs on Llama 3.2 3B (local, Ollama on Core) with a structured FAQ knowledge base built once by a smart model (Claude Opus 4.7).

Chatbot Personality:

  • Empathetic — Acknowledge the frustration of an unpaid debt. "I understand how stressful unpaid invoices can be."
  • Kind but firm — Supportive without being a pushover. Never apologetic for DRE's process or fees.
  • Engaging — Ask leading questions to qualify and move the conversation forward. "How long has this been outstanding?"
  • No promises — Never guarantee a specific outcome, timeline, or recovery amount. Use "typically," "most claims," "can be."
  • Escalate when stuck — If the visitor asks something outside the bot's scope, invite them to submit a claim or contact the team.

Tone examples:

Do Don't
"I understand — unpaid invoices create real pressure." "Sorry about that."
"Most claims at your level resolve within 30 days." "We'll get your money back in 30 days."
"Our process typically recovers 70%+ of claims." "You'll definitely get paid."
"Ready to get started? We can review your case." "You should definitely use us."

PUBLIC knowledge (chatbot can answer):

  • Fee structure & tier percentages
  • Required documents for a claim
  • Recovery timelines per tier
  • Texas construction lien overview (general)
  • LPOA process (what it is, how to sign)
  • Payment/ACH processing info
  • How to submit a claim
  • Contact/support info
  • Link to Fee Calculator
  • Link to Terms of Service & Privacy Policy

PRIVATE knowledge (NEVER in chatbot):

  • AI claim scoring methodology
  • Debtor research / skip tracing techniques
  • Internal weakness analysis criteria
  • Approval workflow and review criteria
  • Specific claim details or case data
  • Partner law firm names or referral terms
  • Internal fee negotiation limits
  • FDCPA/TDCA compliance strategy
  • Any internal team notes or communications

Future Automation (N8n on Monitor Server)

AI Web Scraper Agent — https://n8n.io/workflows/2006-ai-agent-that-can-scrape-webpages/

  • 4.6/5 rating, free workflow template
  • ReAct AI agent that scrapes web pages for skip tracing
  • Use cases: Texas SOS lookups, county court records, debtor asset searches
  • When: Remind Germaine about this when building DRE portal — can offload debtor research to n8n workflows
Reference in New Issue View Git Blame Copy Permalink