Internal development teams and executives are increasingly looking at AI to automate the creation of internal developer platforms. However, cobbling together open-source components with AI is a far cry from building a scalable, secure, and “enterprise-grade” environment. In this conversation, we explore why betting your internal infrastructure on AI-generated platforms is a high-stakes risk and why human-led architecture still dictates the long-term success of a platform.
There’s also an excerpt.
Relative to your interests, Friday afternoon
AI money is absurd again - Anthropic at $380B, Waymo at 360x revenue - while software CEOs get a 2026 “grow up” playbook. Meanwhile, enterprise AI reality stays slow and messy, coding hype surges, and burnout + Ballard remind us what the work rhythm used to look like.
I use the HA HA! BUSINESS! meme a few times each decade. The original is pretty small and pixel’y, so I asked the robot to upscale it and it did a damn fine job:
Related to your interests, Thursday
Europe doubles down on sovereign cloud, Anthropic shakes markets, and AI seeps into meetings, legislation, and IT services. Plus: Java for AI, portable agent “skills,” ERP disdain, and cultural scale reality checks.
Ask yourself: which lock-in would an enterprise CFO prefer: Being locked into a CRM that holds 15 years of customer data, process customizations, and institutional context that would take two years and $50 million to migrate? Or, being locked into a foundation model that could be swapped for a competitor by changing an API endpoint?
You had thought that maybe, oh, because you could be more productive with AI, then you save some time, you can work less. But then really, you don’t work less. You just work the same amount or even more."
🔗 The first signs of burnout are coming from the people who embrace AI the most
In its latest set of predictions, First said that this year, the upper bounds of its 90% confidence interval in fact approaches 118,000 CVEs, and according to the data, realistic scenarios suggest 70,000 to 100,000 disclosed vulnerabilities are “entirely possible”. The median figure for 2026, it said, would most likely be around 59,000."
Enterprise AI needs new apps, enterprise AI doesn't need new platforms
Enterprises should focus on resilience and learning in AI projects by experimenting with applications rather than building new infrastructure.
Relative to your interests, Saturday morning
The Draw Boy - Remove a bottleneck (usually humans), and supply can meet demand. New demand is created, people but more, new things are invented, people but those. New roles are often created to handle the new businesses. It Turns Out That When Waymos Are Stumped, They Get Intervention From Workers in the Philippines - Weird. The off-shoring jobs angle is ponderous. The top 1,000 Roblox creators earned an average of $1.
If everyone can ship software, what will distinguish the successful companies from the apps that are lost in the noise? It’s no longer enough to just spend more time coding, or to be the first with a good idea.
I think it’s going to become even more important to grow other aspects of running a software business:
- Marketing
- Customer support
- Documentation
- Building trust
- Servers (speed)
Of course I’ll be writing a lot of code too, adding features, fixing bugs. But that’s the bare minimum now.