Thomas Aertgeerts

The through-line

Everything Thomas Aertgeerts says circles back to a single diagnostic error he made early and corrected in public: he was selling to the wrong party. Two or three years of building software for accounting and law firms ended with the realisation that the firms did not actually hurt enough to change, while their clients did. "Wij moeten niet die boekhoudkantoren helpen, want die hebben op zich niet zo'n groot probleem. Nee, nee, wij moeten de klanten van boekhoudkantoren helpen, want die zijn allemaal mega ontevreden." [2] That pivot turned a pure technology company into an accountancy firm with a technology DNA [1], and it set the shape of everything that followed: own the service, own the software, and judge the result by whether the small entrepreneur at the end of the chain is better off.

The second, later movement in his thinking is that AI has made this position far cheaper to hold than it was. Capital requirements for building technology have collapsed, to the point where "zeker AI is een game changer. Met zelfs met ene briljante developer kunt gij bij wijze van spreken de wereld veroveren" [1]. Where the argument once had to be made on service quality alone, it is now also an argument about economics: a lean team can build what was impossible two or three years ago [2], and the profession itself is heading somewhere structurally different.

On who actually has the problem

The insight that founded Astro was that dissatisfaction and purchasing power sat on opposite sides of the table. Accounting firms, presented with digitalisation, felt no urgent pain; their clients felt a great deal [2]. Part of that pain is structural and, in his reading, deliberate. Large accounting groups do not refuse small clients such as doctors or consultants outright, they quote €450 to €600 a month so the client either walks away or overpays [1]. His objection is double: "Ik vind dat één niet ethisch, maar ik vind dat vooral ook economisch ongelooflijk dom" [1], stupid because today's small client is tomorrow's large one. He also arms clients with a diagnostic of his own: if a traditional firm promises real-time bookkeeping, that is a red flag, because it is not possible. The honest answer is that traditional setups run at least three months behind, and simply having the books up to date on the spot already gets you eighty percent of the way [1].

On the software facade and the Frankenstein stack

He is blunt about what accounting software actually delivers. "Weet je wat ik echt goed vind aan die software? Dat zijn de reclames. Echt goede reclame. Weet je wat ik echt slecht vind? Al de rest." [2] Providers sell "no hands accounting" while only twenty to thirty percent is genuinely automated; the rest is keyed in manually by junior bookkeepers or offshore workers, because OCR can read an invoice but cannot make an intelligent booking decision [2]. The consequence for the firm is a stack of twenty to thirty tools, each promising seamless integration and underdelivering, taped together into something he describes flatly: "De realiteit is een monster van Frankenstein." [1] Astro's answer is a single integrated platform covering everything from VAT filing to advice to invoicing [1].

On service-as-a-software

The model he defends is a hybrid he calls service-as-a-software: platform and human service so intertwined that the client no longer notices whether AI or a person handled a given task, a category he expects AI to accelerate as services and software companies converge [2]. The commercial argument is that it needs far less capital than pure SaaS, because costs and revenue rise in parallel; you hire accountants only when new clients arrive, and every euro of accountancy revenue goes back into R&D [2]. This is also why he refuses to talk about engineering headcount. Team size for developers is irrelevant; the question is what return each euro invested in technology yields [2].

On AI and the reversed world

His clearest claim about capability is that "die AI is nu beter dan een boekhouder tussen 0 en 5 jaar ervaring" [2]. The organisational consequence is what he calls the reversed world: new interns are trained by studying the AI's reasoning instead of doing low-level work [2]. At the same time, because entry-level bookkeeping is automated, juniors straight out of school take on harder work and sit in client conversations from their first month [1]. As he puts it, "die mensen vandaag die doen complexer werk dan een accountant in een ander kantoor die dat drie of vier jaar bezig zijn" [1], which he treats as a retention weapon in the war for talent. Looking out five years, he expects the accountant to do zero compliance work and instead hold eight advisory calls a day, turning accountancy from an introvert's job into a people profession, with far fewer accountants serving far more clients [2].

On scaling by saying no

Growth to thirty-plus people in three years came from subtraction as much as addition. "Hoe kunt je schalen door klanten te weigeren? Dat is wat we gedaan hebben." [2] Astro deliberately turned away very small sole proprietorships and larger KMOs to concentrate on small companies: consultants, freelancers, doctors, lawyers. Startups and scale-ups were only added once the technology was mature enough to hold service quality [2]. Behind this sits a broader argument about specialisation. Expecting one accountant to master tax law, bookkeeping, business advice, financial ratios and HR is fantasy, and even inside tax law every lawyer occupies a narrow niche: "ene mens, hoe briljant dat je ook zijt, kan nooit in al die aspecten uitblinken" [1]. Firms that claim to serve everyone from students to large enterprises are therefore making a promise they cannot keep [1]. The result of the focused version is a client base with an NPS above seventy [2], and pricing that he himself finds startling: "100 tot €200 op 10 of 15 minuten. Dat is toch crazy, per maand hè." [2]

On people, feedback and knowledge that outlives them

The trait he screens hardest for in hiring is how someone handles feedback, while conceding that interviews cannot test it, because everyone gives socially desirable answers and the real signal only appears under stress [2]. Internally he enforces a culture of parler vrai: saying hard things transparently and respectfully [2]. He also refuses to let expertise live in heads. Knowledge belongs in the organisation, so Astro writes playbooks that let any employee answer basic optimisation questions, on the assumption that people will eventually leave no matter how good the environment is [1]. The same pragmatism governs co-founding with family: make the agreements while things are going well, then "put them in a drawer and hope dust gathers on them" [2].

On investors as compressed time

He counts switching investors among his top lessons. The first was a corporate backer, an accounting firm that resisted the very ideas Astro stood for; the replacement was a serial entrepreneur [1]. The value he places on the second is measured in time rather than money: one hour of conversation with the right investor can fill three months of management implementation with ideas and inspiration [1], and the founders progress roughly three months faster for each such hour, because their mentor has already met every operational problem in another company. It is knowledge, he says, you cannot put a euro value on [2].

On the founder's temperament

He is unsentimental about what the job is: "Als je ondernemer zijt, alles is uw probleem. Alles, alles." [2] His counterweight is composure, summed up as "kalmte alleen kan u redden" [2], and a self-check against the founder's habit of measuring today against an unreachable ideal: every year, ask whether you would have signed twelve months ago for the position you hold now [2]. Curiosity is the engine underneath it, stated plainly as "ik wil leren, leren, leren, leren, leren" [1]. He left corporate safety at an unpromising moment, his last day inside it being the first day of the COVID lockdown [2], and reports no second thoughts on either count: "ik heb nog op geen enkele moment, ook niet als het moeilijk ging, gedacht van was ik nog maar advocaat gebleven. Nul keren" [2], and "we zijn nu zes jaar verder, nog geen minuut spijt van gehad" [1].

Takeaways

• Check which side of the transaction feels the pain before you build: accounting firms were comfortable enough to ignore digitalisation, their clients were not, and that reversal turned a software company into an accountancy firm with technology DNA [1][2].
• Pricing a client out is not just unfriendly, it is bad business, because the small client quoted €450 to €600 a month today is the large client someone else keeps tomorrow [1].
• Treat "no hands accounting" claims sceptically: only twenty to thirty percent is truly automated, the rest is manual keying, and OCR can read an invoice without being able to book it intelligently [2].
• A promise of real-time bookkeeping from a traditional firm is a red flag; three months behind is the honest baseline, and books that are simply up to date on the spot already deliver eighty percent of the benefit [1].
• Refusing clients is a scaling tactic: exclude the segments your technology cannot yet serve well, and only widen the funnel when the platform can hold the service level [2].
• Judge engineering by return per euro invested in technology, not by headcount, since AI now lets a very small team do what was impossible two or three years ago [1][2].
• Write playbooks so basic client questions can be answered by anyone, on the assumption that every good employee eventually leaves [1].
• Hire on how people take feedback, and accept that interviews will not reveal it because everyone answers as expected under no pressure [2].
• Use the twelve-month test yearly: would you have signed a year ago for where you stand today? It is the antidote to comparing yourself with an ideal you will never reach [2].