The challenge
Afinoz's consumer-lending pipeline mixes inbound applications from multiple partner channels with widely varying document quality, and underwriter prep is a manual sift today. Decisioning time stretches past three days on edge cases, and there's no agent layer parsing documents or building underwriter packets.
How we approached it
Ajuni is scoping a pilot with the Afinoz product team where Lakshya handles application triage and document parsing, and Veer attaches RBI-aligned explainability to every recommendation. The pilot will run on Azure India inside Afinoz's tenant, with the underwriter staying in the loop on every decision.
What the pilot will prove
- Pilot will prove sub-3-day decisioning on triaged apps
- Pilot will prove RBI-aligned explainability per recommendation
- In scope: document parsing across partner channels
- In scope: underwriter prep packets auto-assembled
- Scoping target: full pipeline coverage
Stack & guardrails
Integration & deployment
- Azure India · Afinoz tenant
- Lending platform integration
- Document OCR + parse
- Claude Sonnet 4 · API
- Explainability audit store
- Partner-channel webhook ingest
Compliance & audit
- RBI explainability
- DPDP Act 2023
- ISO 27001
Timeline
-
Week 0–1
Product-team scoping
Triage criteria and explainability surface designed with product and risk.
-
Week 2–3
Document parsing build
Parser tuned against partner-channel document variability.
-
Week 4–5
Shadow triage
Lakshya triages live applications alongside underwriters; outputs compared.
-
Week 6
Pilot review
Pilot outcomes reviewed with risk team; rollout decision finalised.
