Piercing the Corporate Veil: How We Built an Autonomous Debtor Syndicate Graph with TigerGraph & Next.js

At Hacker House Goa 2026, our team built RevClaim AI—an autonomous voice recovery copilot and zero-leakage MoR settlement engine designed to resolve India's ₹8.1 Trillion B2B delayed payment crisis.

While autonomous voice agents handle high-EQ debtor communication, one fundamental commercial challenge remained: Serial Defaulter Rings.

1. The Problem: The Corporate Veil Shield

In commercial debt collection, chronic defaulting companies rarely operate as isolated entities. Unscrupulous founders frequently abandon shell entities with accumulated debts, while simultaneously operating profitable sister companies under shared Director Identification Numbers (DINs) registered with the Ministry of Corporate Affairs (MCA).

In a traditional relational SQL database, uncovering linked defaulter rings requires recursive 4-way self-joins across companies, directors, and default records. At scale, this triggers an $O(N^k)$ Cartesian explosion that leads to production query timeouts.

2. Our Solution: TigerGraph Savanna GSQL Engine

To pierce this corporate veil in real time, we architected DebtorSyndicateGraph using TigerGraph Savanna.

Graph Schema Topology

• Vertex Types:
  • DebtorCompany: Represents registered commercial entities (CIN, GSTIN, legal name).
  • Director: Represents individual board members and beneficial owners (DIN, active MSME cases).
  • DefaultRecord: Holds verified outstanding debt exposure and statutory filings.
• Edge Types:
  • DIRECTED_BY: Undirected/Bi-directional relationship connecting companies to directors.
  • HAS_DEFAULT_RECORD: Directed edge from corporate entities to default records.

3. The Multi-Hop GSQL Traversal (detect_defaulter_syndicate)

We wrote an optimized GSQL query that executes a 3-hop traversal in under 12 milliseconds:

CREATE QUERY detect_defaulter_syndicate(VERTEX<DebtorCompany> target_company) FOR GRAPH DebtorSyndicateGraph {
    SumAccum<DOUBLE> @@total_linked_defaults = 0;
    SumAccum<INT> @@linked_entities_count = 0;
    SumAccum<INT> @@msme_violations_count = 0;
    MaxAccum<DOUBLE> @@highest_syndicate_risk = 0;
    SetAccum<STRING> @@sister_company_names;
    SetAccum<STRING> @@directors_involved;

    Start = { target_company };

    // Hop 1: Target Company -> Shared Directors
    Directors = SELECT d FROM Start:s -(DIRECTED_BY:e)-> Director:d
                ACCUM @@directors_involved += d.full_name,
                      @@msme_violations_count += d.msme_cases_count,
                      @@total_linked_defaults += d.total_defaulted_across_entities;

    // Hop 2: Directors -> Affiliated Sister Entities
    SisterCompanies = SELECT c FROM Directors:d -(<DIRECTED_BY:e)- DebtorCompany:c
                      WHERE c != target_company
                      ACCUM @@sister_company_names += c.legal_name,
                            @@linked_entities_count += 1;

    // Hop 3: Sister Entities -> Default Filings
    Defaults = SELECT def FROM SisterCompanies:c -(HAS_DEFAULT_RECORD:e)-> DefaultRecord:def
               ACCUM @@total_linked_defaults += def.total_unpaid_amount;

    // Compute Dynamic Graph Syndicate Risk Score (0 - 100)
    @@highest_syndicate_risk = 25.0 + (@@linked_entities_count * 20.0) + (@@msme_violations_count * 15.0);
    IF @@highest_syndicate_risk > 99.0 THEN
        @@highest_syndicate_risk = 99.0;
    END;

    PRINT target_company.legal_name AS target_company,
          @@highest_syndicate_risk AS syndicate_risk_score,
          @@sister_company_names AS affiliated_companies,
          @@total_linked_defaults AS total_syndicate_exposure_inr;
}

4. End-to-End Autonomous Workflow

1. Intake: Freelancers ingest overdue invoices into the RevClaim Cockpit.
2. Graph Intelligence: TigerGraph analyzes debtor CIN/DIN in <12ms, surfacing cross-entity liability (e.g. ₹4,65,000 exposure across 2 sister shells with an 87/100 risk score).
3. Voice Negotiation: The autonomous voice specialist (Sarah AI) references the statutory MSMED Act Sec 16 interest shield and corporate syndicate exposure during live phone calls.
4. Instant Settlement: Debtors settle immediately via Dodo Payments Merchant of Record (MoR) hosted links.

5. Results & Conclusion

TigerGraph transformed what would have been a complex, manual fraud investigation into an automated, sub-second pre-call intelligence loop. By uncovering shared director liability before making a single phone call, freelancers regain bargaining power and recover overdue receivables faster.

• GitHub Repository: https://github.com/Niss54/RevClaim-AI
• Live Platform: https://rev-claim-ai.vercel.app