Platform engineering for the AI era
Field notes on platform engineering for the AI era. DORA 2025 analysis, the three persona platform user, and the Foundations Framework in practice.
Most DORA implementations produce numbers that look authoritative but cannot survive a single definitional question. Mat Caniglia explains the four failure modes behind unreliable engineering metrics and what Signal Integrity actually requires.
How much cognitive load is your platform absorbing for your developers, and how do you know? Most engineering metric frameworks cannot answer that question. This post explains why, and what to instrument instead.
DORA 2025 calls AI an amplifier, not an accelerator. It magnifies what your platform already does, good or bad. Mat Caniglia explains the three observable signals that tell you which side of that curve you are on, and why the METR paradox is an argument for better platforms, not fewer AI tools.
Most DORA implementations produce numbers that look authoritative but cannot survive a single definitional question. Mat Caniglia explains the four failure modes behind unreliable engineering metrics and what Signal Integrity actually requires.
Most golden paths fail the moment a deadline hits. The path the team uses under pressure is the real path. Here is how to build one they choose instead of one they tolerate.
How much cognitive load is your platform absorbing for your developers, and how do you know? Most engineering metric frameworks cannot answer that question. This post explains why, and what to instrument instead.
DORA 2025 calls AI an amplifier, not an accelerator. It magnifies what your platform already does, good or bad. Mat Caniglia explains the three observable signals that tell you which side of that curve you are on, and why the METR paradox is an argument for better platforms, not fewer AI tools.
Most internal developer platforms fail before they reach meaningful adoption. After years of field audits, three patterns explain nearly every case.
Cognitive Load is the metric platform engineering quotes most often. Most teams measure it wrong, then act on the wrong signal. Mat Caniglia walks through what Skelton and Pais nailed, where the metric stops being useful, and what comes after.
The DORA AI mirror effect. Platforms with strong delivery foundations gain code quality when AI is adopted. The 2024 DORA report measured a 7.2 percent stability decrease on weak platforms. Mat Caniglia explains the mechanism, the operational signals to track, and what to fix first.
Cognitive Load measures the symptom. Cognitive Absorption names the platform engineering design discipline that responds to it. Three operational signals platform teams can track. Why shift left turned into shift dump. The 2000 paper that nobody in platform engineering reads.
Token usage is an infrastructure cost, not a productivity metric. Here's why most AI measurement frameworks are broken, and what to track instead.
Most engineering orgs that want an IDP aren't ready for one. Here's how to know which side you're on, and what to fix before you build the portal.
Long-form analysis on platform engineering, DORA, and AI readiness from Mat Caniglia. Sent when there is something worth reading.
Two ways to start
A thirty minute strategy call, or a fifteen minute self diagnostic. Both close with a roadmap.
Want to read first? See the Foundations Framework