On 10 August 2026, TactML reached an important milestone: we internally deployed ABLE, our very own AI accounting assistant, and began using it in our day-to-day accounting work.
ABLE is designed to intelligently automate much of the repetitive work involved in accounting while keeping our accountants firmly in the loop. At the same time, it will give our clients a better way to interact with their finances—allowing them to see new transactions daily, manage vendor bills and customer invoices, communicate with their accountant, and access key financial insights from a modern interface.
In October 2026, we plan to start making ABLE available exclusively to TactML clients at no additional charge to their monthly service fees.
Getting to this point has taken roughly two and a half years. In this post, I’d like to take you behind the scenes of that journey.
How ABLE Works
Before getting into the story, it is worth explaining what we have actually built.
Every day, ABLE ingests new bank transactions from our clients’ local bank accounts. Those transactions are then analysed by HummingBird, our proprietary machine-learning model.
HummingBird assigns each transaction an accounting label together with a confidence score. Where necessary, it can also match transactions to the relevant vendor bill or customer invoice.
ABLE is connected directly to TwinField, the accounting system we use. When HummingBird reaches 100% confidence in a categorisation, the transaction can be posted automatically to TwinField.
Importantly, automation does not remove human oversight. Every transaction is reviewed by an accountant—including transactions that ABLE has been confident enough to post automatically.
Our goal is not simply to remove accountants from the process. It is to remove as much unnecessary administrative work as possible so that our accountants can spend more of their time doing what people are better at: answering questions, solving problems, supporting clients and applying professional judgement.
We expect ABLE to eventually increase the productivity of our accounting team by two to three times. Many activities that previously required navigating through TwinField and completing several manual actions can now require only a few clicks—or, in some cases, no clicks at all.
For our clients, the change should be just as meaningful.
Today, clients primarily interact with TwinField to create and manage sales invoices. ABLE brings those activities together within a modern interface and expands what clients can do. They will be able to:
- see newly imported bank transactions each day;
- upload vendor bills and see which vendors they owe;
- create sales invoices and see which customers owe them;
- view key financial information and insights in one place; and
- leave comments directly on relevant transactions as feedback for accountants
In other words, ABLE is intended to become the everyday interface between our clients, their financial information and our accounting team.
But two and a half years ago, we were nowhere near this point.
The Beginning
Roughly two and a half years ago, TactML applied for EU funding to develop what would eventually become ABLE. At the time, the business felt like a boat drifting in the middle of the sea.
This was before we developed our water-drop logo. It was before we adopted our company philosophy of “one drop at a time fills an ocean.” Back then, I was mainly trying to turn on the funding tap and make a big splash.
Alongside EU funding, we explored microfinancing and had conversations with potential angel investors.
There was only one problem: we still had a fairly vague idea of exactly what we wanted to build.
We did not yet have a sufficiently clear business strategy. We had not settled on the right AI and engineering architecture. Our product roadmap was not mature enough. In hindsight, there were many things we still needed to understand.
Needless to say, the funding tap remained firmly closed and our bucket remained empty.
But every rejection taught us something.
The feedback forced us to explain our ideas more clearly, question our assumptions and think more deeply about what kind of company TactML needed to become before we could build the product we had in mind.
Similar to the financial difficulties my family experienced while I was growing up, it sometimes seemed that cash still ruled everything around me.
Eventually, however, we realised that waiting for a large cheque was not going to build ABLE. We would have to move forward “stap voor stap,” as we say in Dutch—or, as we now say at TactML, one drop at a time.
With faith in God, perseverance and financial support from family, paying customers and business partners, we continued. And over the following two and a half years, TactML changed significantly.
Looking back, I think that growth happened in three areas: character, content and capabilities.
Character
Firstly, the last two and a half years have given me an opportunity to develop my faith in God and my trust in His provision.
Bootstrapping a technology company like TactML has not been easy. There were several points along the journey where running out of money was a very real possibility. Yet each time, a way forward appeared—whether through a large client project, support from family or another form of provision.
Those experiences helped me grow as a leader and entrepreneur.
I have had to learn how to remain tenacious when circumstances were difficult, how to make better use of the tools and resources available to us, and how to create an environment in which my teammates can excel in their own roles.
I am increasingly convinced that one of the most important outcomes of any difficult journey is not simply reaching the destination, but who you become along the way. As one of my favourite speakers, Myron Golden, puts it: we are human beings—not simply human doings or human havings.
I have also had the privilege of seeing that growth in my colleagues. Building ABLE has required patience, curiosity, resilience and a willingness to keep learning. Those qualities have become just as important to our progress as any individual piece of technology.
Content
While we were dreaming about the product we wanted to build, we were also gradually gaining clarity on our mission and values.
As an INTJ, I particularly enjoy developing conceptual frameworks that make complicated ideas easier to understand and remember. Over time, research, reflection and conversations with my colleagues led me to write articles such as Sun, Sand and Sea, Rice and Peas and HummingBird.
On the surface, these are pieces of content. However in practice, they became something more important.
Writing forced us to articulate what we actually believed about our business strategy, agentic accounting, artificial intelligence and the architecture required to bring those ideas together.
The articles became a way of codifying our thinking and communicating it to teammates, partners and clients—and perhaps, one day, investors.
In the beginning, we thought our main problem was finding enough money to build a product.
Over time, I realised that we first needed clarity about what we were building, how we wanted to build it and, most importantly, why TactML should exist in the first place.
That period of thinking and writing helped transform ABLE from a vague product idea into something that could actually be engineered.
Capabilities
The technological environment has changed dramatically.
When we first applied for funding, AI-assisted coding was largely associated with tools that could autocomplete individual lines or small pieces of code.
I had just completed my MSc in Data Science, specialising in finance, in June 2024.
Since then, both the technology available to us and the capabilities within our team have advanced considerably.
Today, our lead software engineer and AI researcher both use AI coding tools such as Claude Code and Codex as part of their development workflows. These tools have significantly changed the speed at which small teams like ours can design, build, test and iterate on software.
But the more important change has been within the team itself.
Across TactML—from accountants to engineers and AI researchers—I have watched our collective knowledge, judgement and technical capability grow. Ideas that would once have required significantly more time and resources to implement can now be developed and tested far more quickly.
At the same time, our accountants bring the practical domain knowledge that an accounting system needs. Our engineers understand how to turn concepts into reliable software. Our AI researchers continue improving the models underneath the product.
ABLE sits at the intersection of all three. That is what makes this internal launch particularly meaningful.
Two and a half years ago, we had an idea for an AI accounting assistant and were looking for someone to fund us so that we could build it.
Today, our own accountants are using it.
Humans and AI Working Together
There is also an important principle behind how we have designed ABLE.
We do not believe that introducing AI into accounting means removing accountants from the equation.
Accounting involves judgement, responsibility and trust.
HummingBird can classify a transaction and tell us how confident it is. ABLE can automate actions and communicate with TwinField. But our accountants remain responsible for reviewing the underlying accounting work.
The objective is therefore not automation instead of people. It is automation in service of people.
If ABLE can eliminate repetitive clicks, automatically prepare routine work and bring the information our team needs into one place, our accountants have more time available for the work that creates greater value for clients.
That might mean answering a question more quickly, spotting something unusual, helping a business owner understand their numbers or simply being more available when a client needs support.
To me, that is where the real value of AI in accounting begins.
One Drop at a Time
So, with an abundance of gratitude, I would like to mark this milestone in TactML’s journey by thanking God, my family, my co-founders, our colleagues, clients and business partners.
ABLE was not built because one large investment suddenly changed everything.
It was built gradually.
One lesson learned from a rejected funding application.
One client project.
One article.
One model improvement.
One engineering decision.
One transaction.
One line of code.
Every answered prayer, encouragement given, piece of wisdom imparted, transaction booked and line of code written created a ripple effect that helped bring us to this point.
And there is still a long way to go.
Our internal launch on 10 August is not the destination. Neither is making ABLE available to our existing clients in October. They are milestones on a much longer journey toward the kind of accounting firm we believe TactML can become.
Two and a half years ago, we were hoping someone would turn on the funding tap.
Instead, together, we learned to fill the bucket ourselves.
One drop at a time.
