Why process automation is the first AI win for banks and lenders
Article
Where should a bank or lender put AI first? We asked the leaders who work with AI inside banks and lenders in the Philippines and Indonesia where it could create the most value over the next 12 months. In both markets, process automation came first. This article covers why automation leads, what it usually means in a bank or lender, the step a lot of teams skip before automating and what it takes to get from a working test to daily operations.
Only got five minutes? Here are the key takeaways of this article:
Process automation was the most common answer in both markets, chosen by 60% of Philippine respondents and 30% of Indonesian respondents.
Automation is where AI’s value is easiest to measure, because operations teams already track time, cost and errors.
Before you automate a process, map how it runs today and agree its best route, otherwise you’ll repeat its problems at speed.
Getting from a working test to daily operations takes current data, a stable way for a model to act inside the workflow and controls that record what it did.
Get the full findings in From AI interest to operational value for the Philippines and From AI adoption to AI at scale for Indonesia. You can also book a call with vanda.jirasek@oradian.com to talk through your first process.
Want to find out more first? Read the rest of this.
Automation came first in both markets
Philippine respondents put process automation first: 60% chose it as the area where AI could create the most value for their institution over the next 12 months. It led at both stages of adoption, chosen by 58% of respondents at an early stage and 63% of those further along. Respondents in risk and compliance, technology, operations and finance all picked it.
Indonesian respondents spread their choices more widely, with no option chosen by more than a third of them. Process automation was still the most common answer (30%), ahead of personalised customer experiences (21%) and better credit decisions (18%).
What automation usually means in a bank or lender
Our survey didn’t ask which processes respondents had in mind, so these examples are ours. In banks and lenders, automation usually targets manual, repeatable work such as loan processing, document checks, customer onboarding, know-your-customer reviews, reconciliations, report preparation, servicing requests and compliance checks.
It makes a strong first business case. Time, cost and error rates are numbers operations teams already track, so the effect of AI shows up in figures the business trusts. Sounds great, right?
The catch is that an automated process that runs end to end usually crosses several systems, from origination and core banking to payments, document stores and reporting. Every handover between them is a place where automation can stall.
Map the process before you automate it
Before automating anything, map how the process runs today, find the steps that add delay or rework and agree the best route through it. It’s an easy step to skip.
Automating the agreed route gives you the full benefit. Automating the process as it stands repeats its problems at speed, from the duplicate check nobody needs to the approval that sits in an inbox for days.
Three things to do first:
Measure: pick three to five high-cost manual processes and record their time, cost and errors now, so you can show what AI changes.
Map: walk each process step by step with every system it touches and agree the best route through it.
Agree: decide early with risk and compliance which automated steps will need a person to review them.
From a working test to daily operations
MIT’s Project NANDA, in its 2025 report The GenAI Divide, found that 95% of organisations deploying generative AI are getting zero return on their investment. In banking, the step from a working test to daily work is where many programmes stall.
Taking automation into daily operations depends on three capabilities:
Current data: operational data that’s available at the moment the process runs.
A way to act: a stable way to call a model or decision engine from inside the workflow.
A record: controls that record what the AI did and why.
Without them, every move into production becomes its own project, with changes to core systems and the operational risk those changes bring.
How Oradian powers automation
Oradian is the AI-native banking platform for banks and lenders in emerging markets. It covers all three:
Current data
Database Access gives your teams a governed, read-only copy of your production database, continuously synced and held on infrastructure you control, so automations and models work from complete, current data without touching the live system.
A way to act
750+ native, versioned API endpoints and real-time events through Kafka and HTTP webhooks let a model act inside the workflow at the moment something happens.
A record
Every product change, partner connection and automated step is logged and ready for regulators. Your own rules run in Custom Code, versioned and auditable.
Oradian also builds real-time reporting dashboards with an AI agent your teams can prompt, so you can see what an automation changes as it happens.
Esquire Financing in the Philippines went live on Oradian in three months and moved its loan approvals from days to hours.
Start with one process
Automation came first for a reason: the value is easy to see and the processes are already yours. Pick one, map it and automate the best route through it. Read the full findings in From AI interest to operational value and From AI adoption to AI at scale, then book a call by emailing vanda.jirasek@oradian.com to choose your first process before the end of the year.