Myriam Broeders

The through-line

The single argument running through everything Broeders says is that AI is no longer something a company chooses to adopt. The capability arrived in the tools whether or not anyone asked for it, so the only real variable left is mindset. "Every company is a tech company... en zo is ook nu every company becoming an AI company, omdat er juist zodanig veel ingebakken zit." [1][2] The corollary is blunt: firms that lack an AI-first mindset and fail to bring their people with them will struggle regardless of what sector they are in, and some will not make it. "Dat is wel een evolutie waar dat er een aantal bedrijven gaan zijn die het heel lastig gaan hebben om te overleven." [1][2]

What tempers that urgency is her insistence that the response should be paced rather than dramatic. She is not arguing for a leap into disruptive reinvention, but for a staged build-up in which productivity comes before teams, teams before core processes, and core processes before anything radical [1][2]. She also declines the pessimistic reading of the same trend: the work changes, the volume of work does not shrink [1][2].

On why adoption only started now

The technology is not new; the usability is. Companies could already experiment with AI components such as speech-to-text and image recognition from around 2016, and Microsoft was pushing "democratization of AI" in 2016-17, but adoption barely got off the ground [1][2]. What changed with generative AI is that the capabilities became out-of-the-box and, in her framing, unavoidable [1][2]. That distinction underpins the whole AI-first case: opting out stopped being an option once the intelligence was baked into the tools people already open every morning [1][2].

On building AI-first in layers

Her adoption model is explicitly a marathon-training analogy rather than a sprint [1]. Start with out-of-the-box tools such as ChatGPT and Copilot for individual productivity. Move next to collective intelligence, with agents operating at team level. Then embed AI into core business processes. Only after those layers are in place does attempting disruptive innovation make sense [1][2]. The ordering matters as much as the content: each stage builds the organisational muscle the next one assumes.

On agents and the stages of autonomy

The same staged logic governs how she expects agents to enter the enterprise. First a human working with an assistant, the copilot pattern. Then human-agent teams. Then agent-operated processes, where agents take decisions autonomously on the basis of history [1][2]. Even at that last stage the human remains ultimately in the loop and in charge [1][2]. The tone around all this is one of appetite rather than caution: "Het is een enorm speeltuin, speelgoed, mogelijkheden, sky is the limit." [2]

On transforming while performing

She is candid that the hard part is not the technology but the simultaneity. "We noemen dat dan transforming while performing, om die balans te vinden tussen AI en capabilities te introduceren en toch de gewone business te laten draaien." [1][2] Notably, she does not exempt her own employer. Microsoft's own innovation organisation feels the same tension between introducing AI internally and keeping the regular business running, to the point that she has gone outside for advice, consulting Gartner analysts on how other large players manage the balance [1][2]. That admission is the useful part of the message for anyone assuming the problem is theirs alone.

On jobs and what happens to work

She accepts the disruption at the level of individual roles and rejects it at the level of aggregate work. "Er gaan inderdaad jobs zijn die niet meer zullen bestaan zoals vandaag." [2] Logistics planning is her example of a job whose content will look completely different once agents are involved [1][2]. But she declines the shrinking-pie conclusion outright: "Ik zie niet een toekomst waar dat er minder werk zal zijn voor mensen om aan de slag te gaan en minder mogelijkheden, mindere jobs." [1] Her reasoning is historical, that technology has tended to create new possibilities rather than reduce the amount of work available [1].

On saying yes

Alongside the corporate argument sits a personal one about how opportunity actually arrives. "Ik ben ook iemand die heel veel ja zegt op allerlei zotte ideeën. Dat heeft mij al heel veel goede zaken gebracht, zowel op vlak van privé als op vlak van werk." [1] It reads as the individual-scale version of the AI-first mindset she prescribes to organisations, and it is of a piece with her assessment of where she has ended up: "Ik heb de mooiste job van mijn Microsoft, denk ik wel." [1]

Takeaways

• Treat AI adoption as marathon training, not a sprint: individual productivity with out-of-the-box tools first, then team-level agents, then AI embedded in core processes, and only then disruptive innovation [1][2].
• Survival is a question of mindset, not sector. Companies without an AI-first mentality, and without their people brought along, will find it very hard to survive [1][2].
• Expect agent autonomy in three stages: human with assistant, human-agent teams, then agent-operated processes where agents decide from history, with humans still ultimately in charge [1][2].
• The bottleneck was never the technology. AI components were available to experiment with from around 2016, but adoption only moved when generative AI made the capabilities out-of-the-box [1][2].
• "Transforming while performing" is the real difficulty, and even Microsoft's own innovation organisation struggles with it enough to seek outside advice from Gartner on how peers cope [1][2].
• Individual jobs will change beyond recognition, logistics planning among them, but expect no net reduction in work or opportunity [1][2].