Case study · Lending / Financial Services

Automating complex lending operations with agent-driven workflows.

An AI-orchestrated operations layer that runs Lendly’s data ingestion, reconciliation, CRM updates and customer engagement on schedule, with human approval where it matters.

Discuss a similar build All case studies

Client: Lendly
Industry: Lending / Financial Services
Delivery: Workflow automation, AI agents, CRM and data integrations
Capability: AI Agents & Automation

The challenge

What Lendly was up against.

Lendly’s critical workflows were spread across Salesforce, email and document ingestion, financial transaction files, enrichment tools and review management. The challenge was not building one automation but orchestrating many interdependent workflows reliably.

Disconnected tools and manual hand-offs
Error-prone daily data handling
Hard to audit or scale
Dependent on daily human intervention

The solution

What we built.

Skugal built a modular set of production-grade agents and workflows. Each one runs on a schedule, validates data, detects missing or duplicate records, triggers alerts or approvals, and safely updates downstream systems.

Registration search & state-wise aggregation

Queries Salesforce registration records, groups and merges them by state, removes duplicates and prepares structured inputs for browser automation.

Financial file processing & validation

Pulls daily transaction files from email, stores them in Dropbox and AWS, filters irrelevant records, calculates GST and gross values, and alerts on missing history.

Monthly gross value reporting

Aggregates records by month, updates financial tables, marks processed rows and runs idempotently so re-runs are always safe.

Employee & account assignment

Distributes eligible Salesforce accounts across employees within defined limits and writes assignment status back to the CRM.

Connection verification & enrichment

Orchestrates asynchronous PhantomBuster jobs with secure session handling, polling and result parsing.

Google reviews with AI-drafted replies

Collects and deduplicates reviews, drafts replies with AI and routes every reply for approval before anything is published.

How it works

From input to outcome.

Scheduled triggers
Data fetch & validation agents
Transformation & calculation agents
Duplicate / missing record detection
CRM & table updates
Alerts, approvals & publishing

Business impact

What changed.

Lendly moved from human-driven back-office processes to a structured, agent-driven automation layer with predictable daily execution, safer financial data handling and controlled AI usage with human approvals.

Technology