AI training for banks and insurers in the Philippines
AI training for banks and insurers works when it is built around one real task the team already does, with a before and after number attached. On-Off Group has run workshops for Philippine financial services since 2015, including AIA, AXA, BPI, Security Bank, Metrobank, Sun Life and Insular Life.
This is for heads of digital, CX, operations and learning at Philippine banks and insurers who have been asked to "do something about AI" and want the training to show up in the work, not in a feedback form.
Two things it will give you. A way to judge whether AI training for banks is any good before you book it. And a way to prove afterwards that something changed, so the answer to "did it work?" is a number rather than a shrug.
On-Off Group has run workshops for Philippine financial services since 2015. Cleared client names include AIA, AXA, BPI, Security Bank, Metrobank, EastWest Ageas, Sun Life and Insular Life. Some of the people in the Sun Life sessions in 2015 are now senior across the industry.
The room fills up and nothing changes on Monday
The normal pattern looks fine from the outside. Most people invited book a slot, most of them show up, the feedback score comes back high, and a certificate goes into the HR record. Everyone signs off and moves on.
Then four weeks later someone asks what the training changed. Nobody can answer, because nobody picked a task the training was meant to speed up. People learnt what a large language model is and how to write a better prompt in general, went back to their queue, and carried on the way they always did.
The test is the first real task after the room empties. Did anyone open the AI tool that week, and did it help? If the answer is no, the score was measuring attendance.
So pick the task before you pick the course. One task, one team, something they already do every day: drafting a customer reply, summarising a claim file, writing the first version of a product FAQ, cleaning up branch feedback into something readable.
Train one task, not everyone at once
The common move is to buy licences for the whole department, run one big session, and hope. Usage drops off in a month and the renewal conversation gets awkward. We have written about this in Instead of buying AI licences for everyone, train one real task first.
Better: one team, one task, prompts written on their own live work during the session, and everyone leaves with a prompt they saved and can reuse. Not a template from the trainer. Their own.
That gives you two useful things. First, proof it works in your controls and your tone of voice, which is the part compliance will ask about. Second, a team who can show colleagues how they did it, which spreads further than one big session for everyone.
The topics covered are listed at onoffgroup.com if you want to see them before you shortlist.
Write the before number down first
Most teams have no baseline. That is how a project ends up as a slide pack with nothing to compare against. Nobody wrote down where things stood on day one, so any claim afterwards sounds like marketing.
So before the session, agree one number and write it in an email. Turnaround time on the onboarding flow. Average handling time on a queue. Number of customer replies drafted per person per day. Completion rate on the step where people drop off. Whatever your team already reports, so you are not building a new report to prove a point.
Then measure the same number four to six weeks after. Small honest numbers are fine. A few minutes saved per case, on a queue you already report on weekly, is a real result and it survives being repeated to your boss.
For comparison, on a series of five-day design sprints with a large insurer, customer-facing journeys that used to take six months or more were designed, tested and live in four weeks, and completion rates on the new journeys rose by 80 per cent. That number only exists because the client measured the old journey first.
Who is in the room, and what language they work in
A lot of AI training is written for head office, and the examples make no sense to branch staff. The examples are startup examples. The pace assumes people are comfortable typing paragraphs into a tool in English.
Branch staff, claims processors and contact centre agents handle the customer, so a decent prompt saves them the most time. So the material has to meet them. AI Prompting Essentials on On-Off Academy is in English and Tagalog, with audio narration, for exactly that reason.
Level matters too. A head of department does not need to write prompts all day; they need to judge whether a team's AI output is safe to send to a customer. Those are different sessions with different exercises, even when the tool is the same.
If the step already confuses customers, AI makes it confusing faster
If a step already confuses customers, adding AI to it makes the confusion faster. We see this most often on onboarding and claims, where the form asks for things the customer cannot easily give.
So before you automate a step, watch five real customers try it and note where they stop. Fix the obvious breaks, then apply AI to the part that is genuinely slow. This is why AI training often runs alongside work on the customer journey: design thinking training for bank teams covers the sessions where product, operations and risk people work on the same journey in one room.
What to do this week
- Name one task, in one team, that you would want AI used on first.
- Write down today's number for that task in an email to yourself, with the date.
- Ask the people who do the task every day which part of it they would happily hand over.
- Check that any training you shortlist uses your work as the exercise, not generic examples.
- Book the follow-up measurement for four to six weeks after the session, before the session happens.
More on AI training
- What is corporate AI training and what does it cover
- What AI training for a team costs and how long it takes
- Instead of buying AI licences for everyone, train one real task first
- Design thinking training for cross-functional bank teams
- What UX training covers and who in a bank should take it
- Everything on AI training
Questions people ask
Have you trained people in Philippine banking and insurance before?
Yes. On-Off Group has run workshops for Philippine financial services since 2015, with cleared client names including AIA, AXA, BPI, Security Bank, Metrobank, EastWest Ageas, Sun Life and Insular Life.
How do we know if AI training worked?
Pick one task, write down the current number before the training, then measure the same number four to six weeks after. Turnaround time on a customer journey step is a good one because it is already reported.
Does the training work for branch and operations staff, not just head office?
Yes, as long as the examples are their own work. AI Prompting Essentials on On-Off Academy is available in English and Tagalog with audio narration for exactly this reason.
What should we train first if the budget is small?
Train one real task for one team rather than buying licences for everyone. A small group that uses AI on Monday is better evidence than a full room with good feedback scores.
On-Off Group trains teams, tests products with real customers, finds where a transformation has stalled and builds what gets it moving, for banks, insurers and enterprises in the Philippines, since 2015. How we help with ai training.
