Closing Asia's insurance gaps: systems that were never built to meet.
Three things stop insurers, their sales channels and their repair networks working as one system. Which ones costs most, which ones you cannot see, and what we have learned putting a claims operation live one insurer at a time.
Almost every insurance operation I work with in Vietnam runs across companies it does not control. A carrier sells through channels it does not own. A motor claim is settled in a repair network it does not employ. And yet, a regulator still expects a clear account of the activities across all ecosystems.
Every one of those relationships work. I want to say that first, because the instinct when something is slow is to look for an individual who is ‘lagging’ behind. In my experience there is no such person obstructing efficiency. The truth is something most would either refuse to accept or not think of, and that is how the connections were made in the first place: one at a time, each one sensible on the day it was built.
Three things follow from that. I have put them in the order that they cost you money, which is not necessarily the order you will notice them in.
Connecting the next partner costs what the last one cost
Bringing on a new carrier, a new sales channel or a new repair garage is a big project. The first one earns its budget. People negotiate, discuss terms, conditions and working parameters. It’s important work and needs to be done. This doesn’t need to change.
What I would change is what happens next. The second connection costs about what the first did, because nothing from the first was written down in a form that the second could use. The work may be real but it was also unrepeatable.
An operation that adds partners faster than it retires them carries every connection it has ever built, for as long as it keeps them. That is not a failure of anybody's engineering. It is simply what five years of buying integration one relationship at a time adds up to.
Two partners can be fully connected and still disagree about what was covered.
Somebody rebuilds the story every time it is asked for
When a regulator, an auditor or a partner asks what happened to one claim, somebody assembles the answer manually. They collect and piece it together from several systems. Each holds part of the sequence but not the order.
The answer is usually right. It is the assembling that costs days, people, and a quiet dependence on what those systems still remember. I have watched good teams do this well, but doing it well is still the most expensive way to be correct.
The same word means different things on each side
Inside a company a policy has one shape. Everyone who touches it agrees what a cover is, which claim it belongs to, and which reference number is the real one.
Across a boundary it has as many shapes as there are companies involved, so every pair of systems needs a translation between them. Somebody writes that translation, and somebody else maintains it from then on.
The translation is not the expensive part. The expensive part is that the connection is where the meaning gets settled, and the meaning sits inside the connection rather than in anything either side owns. This third problem is the one you cannot see from inside a single company, and it sits underneath the other two. The connection always costs the same because the meaning is renegotiated every time. The story has to be rebuilt because the parties never agreed what the events meant.
An open standard moves the meaning out of the connection
If the meaning is the problem, the fix is a definition of a policy, a claim and a cover that both sides hold rather than negotiate. That is what an open insurance standard is for. We build on OPIN, which supplies exactly that shared vocabulary. We author the profile each market needs on top of it, closing the gaps a shared standard leaves when it meets one country's rules.
Vietnam is the first.
Something changed there this August, and it changes what I am able to promise you. OPIN was created by the Open Insurance Initiative, published between 2018 and 2022, and then it stopped. Nothing has been revised in four years, and twenty structural problems in it have no route to a fix. We were building on a standard nobody was maintaining. So we took it on: we carry it forward now, publish it under the same licence, and credit the initiative as its origin.
I want to be plain about why it has to be open rather than ours, and it is a harder sentence to write now when we are doing the writing.
Every vendor in this market will ask a carrier to adopt its model. The difference worth caring about is whether the carrier still owns what it built afterwards. A standard that one company controls and withholds is an interface with a marketing name. Adopt it and you take on a dependency rather than gain a model. Open is the only version of this a carrier should accept from a supplier, including from us.
That is easy to say but it needs a receipt. So here is ours. The standard is published at github.com/trisilva/opin under MPL 2.0, and that licence cannot be revoked. Its known gaps are written into it. The twenty structural problems are published as a list rather than quietly patched.
Every accepted change records its reason against the defect it closes, so you can audit the reasoning instead of trusting us for it. And because one company maintaining it is the position I just warned you about, the condition for opening editorship is written into the governance file: two independent implementers, or one sustained outside contributor. There are none today.
What we have learned putting it live
We are doing this in production rather than in theory, one insurer at a time. ClaimFlow, our claims platform, is live in the VinFast service-centre network in Vietnam. PJICO was the first carrier to adopt it, from 2 August 2026, and PVI is in acceptance testing behind them. The panel we are working toward is ten Vietnamese carriers. The network itself runs to more than 420 service centres, and our target for reaching all of them is the end of September, which is a target rather than a result.
I am giving you dates rather than percentages deliberately. It is simply too early. One carrier is live and the second will be soon so any efficiency figure I quoted today would be merely be a projection wearing a number's face. What I will say is that the second carrier is a different kind of work from the first, and that the difference is the whole argument I made above.
What changes, and what waiting costs
When the model is shared and open, two things change in your own operation. Connecting the next partner becomes a configuration rather than a project, so it lands in weeks of somebody's attention rather than a budget line to be fussed over. And answering what happened to a claim becomes a query against a record that was written as the work happened, rather than a tedious, manual reconstruction.
The reason I would not wait is that connections compound. Every partner added on the old terms is one more to migrate later, and the cost of moving rises with each one. This is one of the few decisions in an insurance operation where the price of deferring goes up rather than staying flat.
There is a larger thing beneath it though.
Whoever holds the shared model for this market holds a real position, and that is exactly why it should not belong to one company, ours included.
We hold it today because nobody else was going to, and the test of whether we meant any of this is whether we are still the only ones holding it in five years. An open standard is the only fair arrangement where the carriers, channels and networks each keep what they build while being able to integrate and scale seamlessly.
That is what I would like us to be judged on, more than any platform we ship.