The real story from an AI coding conference is not about robots replacing developers. It is about who finally gets a seat at the table, and why the old gatekeepers are losing their grip.
What I took away is not a prediction. It is more of a reframe. The question is not whether AI will change the job market. It obviously will. The more interesting question is who it changes things for, and in which direction.
The answer that kept coming back to me: the people who have historically been locked out. And at the top of that list, women.
Will AI replace software developers?
Short answer: not the ones paying attention.
The longer version is genuinely complicated. The role of Junior developers got some airtime at this conference, and not in a reassuring way. Some argue that entry-level roles will shrink because AI can now handle the scaffolding work that used to train those people. Others argue the opposite: that juniors who embrace AI tooling are actually more valuable right now, because they arrive without years of habits to unlearn and with a genuine appetite for the new workflow.
Both arguments are probably true in different contexts. What seems less debatable is that the shape of the job is changing faster than job descriptions are. About 80% of enterprise AI projects reportedly fail, and the most cited reason is not the technology. It is people inside organisations who are not willing to leave their comfort zone, including, maybe especially, in leadership positions. The bottleneck is human, not algorithmic.
So the developer who learns to work with AI is not being replaced. They are being promoted past everyone who refused to adjust.
Can you build real things with AI if you are not a developer?
This is the one that kept me up a little.
Vibe coding, the practice of building working software through natural language prompts without writing traditional code, is not a gimmick anymore. It is clumsy in places, and I would not hand it a critical infrastructure project tomorrow. But the distance between an idea and a working prototype has collapsed. Significantly.
What surprised me at the conference was who was least threatened by this. The coding bootcamps, the certification platforms, the learn-to-code institutions. You might expect them to be nervous. Most were not. Their argument: understanding the fundamentals still matters, because when the AI gets it wrong, you need to know what wrong looks like. The tool accelerates. It does not replace the need for judgment.
I am not fully convinced, but I am not dismissing it either. I'll keep watching that space and report back.
What most people get wrong
Everyone is framing this as a loss. Job cuts. Skill obsolescence. Industry disruption. That framing is real, but it is not the whole picture, and it is suspiciously centred on people who already had the advantage.
The conversation that landed hardest for me was about women in tech. The argument, and I think it is a strong one, is that AI is one of the most significant equalising forces the industry has ever seen. Women have historically been underrepresented in tech roles, partly because of gatekeeping, partly because of credentials culture, partly because of how networking and sponsorship work in deeply homogenous industries. AI disrupts all three simultaneously.
When the barrier to building something is no longer five years of computer science, but curiosity and the ability to ask the right questions, the landscape changes. The same logic applies to migrants, to people without elite degrees, to people in regions that were never close to a Silicon Valley pipeline.
Globally, women hold around 26% of tech jobs, a number that has barely moved in a decade. The gender pay gap in tech sits somewhere between 16 and 28% depending on the market. Those are stubborn numbers. But when the tools change, the gatekeepers lose some of their leverage.
This is worth sitting with. The technical debt that kept most people on the outside of the digital economy, the years of learning, the access to the right networks, the cost of entry, is being vaporised. Not completely. Not overnight. But faster than any previous shift in computing.
And here is the asymmetry that nobody talks about enough: individuals can adapt faster than organisations. A person can learn a new tool this weekend. A corporation with legacy infrastructure, internal politics, and a middle management layer that is quietly terrified has to navigate all of that before it can even run a proper pilot. That is a structural advantage that currently sits with small teams, independent builders, and people who have nothing to unlearn.
Every structural advantage that kept people out of tech is slowly becoming a software problem, and AI just shipped the patch.
A note on Fleamio
One thing I can speak to directly is what this shift looks like in practice for a small team. At Fleamio, we are a lean operation. We are not a legacy enterprise. We do not have layers of internal politics slowing down every experiment. When a new AI tool lands, we can pick it up, test it against a real problem, and decide within days whether it belongs in the workflow. That agility is not incidental. It is one of the few genuine structural advantages a small team holds over a funded competitor right now. The window where moving fast actually matters is open. We are using it.
What to actually do
- If you are early in your career: do not wait for clarity. The people debating whether to learn AI tooling are already behind the people using it. Pick one tool, build one real thing with it, understand where it breaks.
- If you have been out of tech or never felt like it was for you: this is the most genuinely open window in a generation. The qualification inflation that kept the door closed for years is losing its grip. The entry cost has never been lower.
- If you are inside a large organisation: the 80% failure rate on AI projects is almost certainly downstream of culture, not capability. The question to push internally is not "what tools are we buying" but "what are we actually willing to change about how decisions get made."
- If you are building something independently: your advantage over incumbents is real right now, but it is time-limited. Organisations are slow, but they are not asleep. The window where a small team can move 10x faster than a funded competitor is open. Act accordingly.
- On the pay gap question: if you are a woman, a migrant, or anyone who has been told the cost of entry into tech was too high, it is worth genuinely reassessing that right now. The tools changed. The old map is not reliable anymore.