Coding agents made building easy. Platform teams that build from scratch are 3.5x more likely to lack a strategy.
The New Stack Charlotte Fleming
AI agents made custom platform tools easy to build. But teams that build from scratch are 3.5x more likely to lack a clear strategy.
Based on reporting by The New Stack, Charlotte Fleming — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Build or buy has never been a neat little spreadsheet exercise. The 2026 Future of Platform Engineering report, based on 379 platform professionals, suggests the real tell is not the feature itself but what the choice says about the team behind it. Teams that mostly build from scratch are less likely to have a clear strategy, and they also hit fewer of their platform goals.
AI has sharpened that old argument. Agents can turn something that once took months into a one-day project, which is a lovely way to save time until someone has to own the thing. The code may arrive faster, but the maintenance, upgrades, and governance do not vanish. If anything, more change now flows through the pipeline.
Most teams still buy. An underlying tool is the primary approach for 51.3% of respondents, ahead of building on a template at 31.2% and building from scratch at 17.5%. High performers lean even harder on tools, at 58.8%, while template building and scratch building trail at 47.8% and 33.3%. That looks less like fashion and more like maturity: as platform choices settle, teams seem to standardize instead of reinventing the same plumbing.
The exceptions are the useful part. Local development environments, one-click projects, and ephemeral environments tilt toward templates because they need stack-specific golden paths that off-the-shelf tools often miss. Ephemeral environments matter most here. Teams that have them were much more likely to say AI improved delivery speed and stability, and more than three-quarters of teams reporting an extremely positive AI effect had them. That points to broader platform maturity, not just one feature done well.
The sharpest split comes in strategy. Teams whose dominant method is building from scratch are about 3.5x more likely to say a lack of clear strategy is a problem, at 57.9%, compared with 15.6% for template builders and 16.7% for tool users. The pattern is not “building is bad.” It is “building by default is a red flag.” In a world where agents can spin up a custom incident tool before the question is even fully formed, the harder part is no longer writing the code. It is deciding why that code should exist at all.
My take — AI-written commentary, not fact-checked reporting
Platform teams don’t get points for hand-rolling every bolt just because an agent can do it before lunch. The grown-up move is boring on purpose: buy the standard bits, build only where the company is genuinely weird, and stop confusing motion with strategy. Open source, commercial tools, whatever — the winner is the team that knows why it picked one.
Read more about this at: The New Stack