Use case #0003

Agency hand-off: the data package Hard Bucket AI sends to recovery agencies

The institution is responsible for the conduct of every recovery agent acting on its behalf — whether directly employed or through an outsourced agency. An agency that engages in prohibited collections conduct creates regulatory liability for the institution, not just for the agency. The Hard Bucket Agent AI sends agencies a data package that is simultaneously complete enough to be effective and conduct-bound enough to be compliant — with every prohibited action explicitly stated and every permitted contact documented.

The institution is responsible for the conduct of every recovery agent acting on its behalf — whether directly employed or through an outsourced agency. An agency that engages in prohibited collections conduct creates regulatory liability for the institution, not just for the agency. The Hard Bucket Agent AI sends agencies a data package that is simultaneously complete enough to be effective and conduct-bound enough to be compliant — with every prohibited action explicitly stated and every permitted contact documented.

The institution's liability for agency conduct

The RBI's guidelines on outsourcing of financial services, the Fair Practices Code, and the collections conduct standards all make one principle clear: the institution cannot outsource its compliance obligations. An agency acting on behalf of an institution is subject to the same conduct rules as the institution's own employees — and when an agency violates those rules, the regulatory consequence falls on the institution, not the agency. This is not a theoretical risk. RBI enforcement actions against NBFCs for collections misconduct have included cases where the prohibited conduct was performed by third-party recovery agencies, not internal staff.

The agency hand-off package that the Hard Bucket Agent AI produces is therefore not just a borrower data file. It is a conduct-bound brief that makes the institution's instructions to the agency explicit, documented, and unambiguous — so that there is no question about what the agency was told to do, and no gap in the conduct instruction that an agency could claim created ambiguity about what was prohibited.

"The agency hand-off is not the end of the institution's responsibility — it is the beginning of a new phase of documented, auditable responsibility. The Hard Bucket AI makes every instruction explicit so there is no question about what the institution authorised."

The agency data package: what it contains

Recovery Agency Data Package — Account HL-2024-7741
Suresh Kumar Mehta · DPD 140 · Referred to: [Agency Name] · Nov 14, 2025
Section 1 — Borrower Identity and Contact (verified)
Full NameSuresh Kumar Mehta
Co-borrower / GuarantorMrs Anita Mehta (wife) — guarantor
Registered AddressHouse No. 42, 3rd Cross, Indiranagar, Bengaluru – 560038
Contact Number+91 98XXX XXXXX (verified · DND: Not registered)
EmployerNew employer confirmed (EPFO) — name withheld per DPDP Act data minimisation
Language preferenceKannada and English
Section 2 — Account and Debt Position
Loan account numberHL-2024-7741
Total outstanding (as of referral date)₹46,59,980
Product typeHome Loan · Secured · SARFAESI applicable
NPA classification dateOctober 6, 2025
Section 13(2) notice servedNovember 14, 2025 · 60-day window active until January 12, 2026
SecurityEquitable mortgage — property at registered address · CERSAI confirmed
Section 3 — Contact and Collections History Summary
Borrower has been engaged throughout the collections process — has answered calls, made 4 partial payments totalling ₹42,000, and stated that income disruption due to employer change is the primary reason for non-payment. A restructure review was conducted in September 2025 — EMI holiday was assessed but borrower income at that stage was insufficient to support even reduced payments. Income has since recovered (new employer confirmed via EPFO in October 2025 with monthly salary approximately ₹85,000). Borrower has not filed any legal challenge to the Section 13(2) notice. Assessment: engaged borrower with recovering income — constructive contact more likely to produce resolution than escalated pressure.
Section 4 — Settlement Authority
The agency is authorised to discuss settlement with the borrower within the range of ₹30,00,000 (floor) to ₹38,00,000 (ceiling). The recommended offer is ₹34,50,000. Any settlement agreement reached with the borrower must be confirmed in writing by the institution's authorised officer before it becomes binding. The agency does not have authority to make binding settlement commitments — it may indicate a settlement range and refer any agreement to the institution for formal confirmation within 48 hours.
Section 5 — Conduct Instructions: Permitted and Prohibited
Permitted contact
Phone calls: 8 AM–7 PM only Maximum 2 call attempts per day WhatsApp messages (DND cleared) Residence visit with prior written notice Discuss settlement range as briefed Accept partial payments — issue receipts Contact guarantor (Mrs Anita Mehta only)
Absolutely prohibited — zero tolerance
Contact before 8 AM or after 7 PM Abusive or threatening language Contact employer or workplace Contact any third party other than guarantor Residence visit without prior written notice False claims about legal action Binding settlement commitment without institution approval Collection from caller ID not registered with institution
● Agency engagement under institution's FPC obligations · All agency conduct is institution's responsibility ● Contact log must be submitted weekly · Conduct complaints are institution's liability ● Hard Bucket Agent AI · LendingIQ · Nov 14, 2025

The four-step agency hand-off process

01
Triggered at DPD threshold · Automated

Agency Referral Decision — Account meets agency referral criteria

The Hard Bucket AI triggers an agency referral when an account meets defined criteria: Section 13(2) notice served and 30+ days into the 60-day window with no borrower response, or DPD 90+ on unsecured accounts where internal collections have not produced engagement. The referral decision includes an account suitability assessment — some accounts (those with active court proceedings, those with hardship flags requiring sensitive handling, or those approaching an agreed settlement) are held back from agency referral regardless of DPD age.

02
Post-referral decision · <30 minutes · Automated

Data Package Assembly — Complete borrower and account brief

The Hard Bucket AI assembles the complete agency data package: verified borrower identity and contact details (data minimised per DPDP Act obligations — only what the agency needs for contact and recovery), account position with outstanding and legal status, collections history summary, settlement authority range, and the full conduct instructions. The package is assembled from the LMS, collections CRM, and legal case file — the agency receives a single integrated brief, not a collection of disparate documents.

03
Package transmission · Encrypted channel

Secure Transmission — Data package sent via authorised channel only

The agency data package is transmitted through the institution's authorised agency communication channel — not email, not WhatsApp, not unencrypted file sharing. The DPDP Act and RBI data security guidelines require that borrower data shared with third parties is transmitted securely and that the third party's data handling obligations are contractually specified. The Hard Bucket AI transmits only through channels where the agency's data processing agreement is in place and logs the transmission timestamp and receipt confirmation.

04
Ongoing · Weekly minimum

Agency Performance Monitoring — Contact log review and conduct oversight

The Hard Bucket AI requires weekly contact log submission from the agency and reviews each log for conduct compliance: are all contacts within the 8 AM–7 PM window? Is the contact frequency within permitted limits? Are any prohibited contact types flagged? Any conduct concern identified in the log triggers an immediate flag to the institution's compliance team — and a formal warning or agency relationship review depending on the severity. The institution's regulatory responsibility for agency conduct is only discharged by active oversight, not delegation.

What agencies must not do — and why the package makes this explicit

The prohibited conduct list in the agency package is not a courtesy note — it is a formal instruction with documented delivery. If an agency engages in prohibited conduct, the institution's first line of defence in any regulatory proceeding is that the conduct was explicitly prohibited in writing, in the agency brief, at the time of referral. This does not eliminate the institution's liability entirely — the RBI holds institutions responsible for agency conduct regardless — but it demonstrates that the institution took reasonable steps to prevent the violation and is material to the severity of any regulatory consequence.

Prohibited: Third-Party Contact

No contact with employer, relatives, or neighbours who are not guarantors

The FPC and RBI guidelines are explicit: recovery agents may not contact employers, relatives, neighbours, or any third party who is not a named guarantor in the loan agreement. The borrower's employer address and any family contact information that exists in the institution's records are excluded from the agency data package — the agency receives only the borrower's own contact details and the guarantor's details. Exclusion is structural, not instructional.

→ Enforcement: employer address excluded from package at data level, not just instructed
Prohibited: Out-of-Hours Contact

Zero calls, messages, or visits before 8 AM or after 7 PM

The contact window is 8 AM to 7 PM. This applies to calls, WhatsApp messages, SMS, and field visits. The agency's call platform should have a technical restriction preventing dialling outside this window — but the Hard Bucket AI's conduct brief explicitly states the restriction so there is no ambiguity and no "system error" defence available.

→ Enforcement: conduct log shows timestamp of every contact — out-of-hours flagged automatically
Restricted: Field Visit Protocol

Residence visit requires prior written notice — at least 7 days advance

A field recovery agent may visit the borrower's registered address — but only after prior written notice has been sent, providing the borrower at least 7 days' advance notice of the visit date. Unannounced field visits are prohibited. The written notice requirement is a hard rule — it exists to prevent conduct that the RBI has classified as harassment, and its breach is among the most serious FPC violations in field recovery.

→ Enforcement: field visit authorisation only issued after written notice dispatch confirmed
Required: Contact Log Submission

Weekly contact log submitted to institution — every contact documented

Every contact the agency makes — call, message, field visit — must be logged with date, time, channel, borrower response, and outcome. This log is submitted weekly to the institution and reviewed by the Hard Bucket AI for conduct compliance. An agency that fails to submit logs, or submits incomplete logs, is flagged for compliance review immediately.

→ Enforcement: log submission required weekly · AI review for out-of-window, excessive frequency
5Package sections — identity, account position, history, settlement authority, conduct instructions
8Explicitly prohibited actions — each documented in writing in the agency brief at referral
WeeklyContact log review cadence — AI reviews agency conduct logs and flags violations immediately
AlwaysInstitution retains regulatory responsibility for agency conduct — oversight is non-delegable

The agency hand-off is not the end of the institution's responsibility — it is the beginning of documented, auditable oversight

The institution that sends a borrower data file to a recovery agency and considers its responsibility discharged is the institution that will face a regulatory action when the agency behaves badly. The institution that sends a conduct-bound brief with explicit instructions, transmits it through a documented channel, reviews weekly contact logs, and flags violations immediately is the institution that has discharged its oversight obligation. The Hard Bucket Agent AI makes the difference between those two positions structural rather than aspirational — the conduct instructions are embedded in every package, the log review is automatic, and the documentation of every instruction is maintained in the compliance record from the day of referral.

← Back to Hard Bucket Agent AI