Adaptive System Design: the loop that stays anchored.
You already shipped this. Bring what you know about how it's built, what it's built on, and what you can actually observe — this tool designs the monitored signals, the adaptation rules with a scoped autonomy level on each, and the standing review cadence that keeps automated tuning from quietly drifting away from the theory it's meant to serve. It designs the loop. It is not the loop.
Kolb Cycle
Scoped Autonomy
Two-Track Measurement
Core Tension
What this tool is built to hold in tension
AI Affordance
AI can watch a live learning system continuously, surface patterns no human has time to see, and adjust in real time — the continuous-adaptation phase becomes genuinely operable for a small team.
Named Risk — Drift from Theory
Each automated adjustment is locally reasonable, and a hundred of them later the system optimizes engagement while the learning design's theoretical spine has been quietly tuned away.
Guardrail
Every adaptation rule carries a theory anchor (what learning-science commitment it must not violate), a scoped autonomy level, and a drift indicator. Anything that touches the theoretical spine is propose-only or human-only, and the loop has a standing human review cadence — anchoring is a schedule, not a sentiment.
Human Judgment Domain
What the design is FOR, which patterns mean something, and every adjustment that touches the theoretical core. The system observes and proposes at scale; conceptualization stays with you.
What this tool does not replace: A learning engineer's judgment, or periodic real evaluation (Phase 3) — telemetry is not evidence of learning.
What I can and can't do:I design the adaptation loop — monitored signals, rules with scoped autonomy, review cadence — from what you tell me about your live system and its theoretical core. I don't run the loop, watch real telemetry, or decide whether automated adaptation is right for your context — that stays yours.
Optional — this tool works fully standalone. Pick or create a project only if you want this loop design saved to the shared Design Journey (and, when a prior Enactment or Local Impact cycle exists, offered as pasteable context for the theoretical core below).
Describe What's Already Live
The system, its theory, and what you can see
Three required fields, one optional. Stating the theoretical core in your own words IS the anchor — this tool can only protect commitments you name.
The live/shipping learning experience and how learners actually move through it.
0 / 3000 · minimum 60
The learning-science commitments this design is built on, in your own words — same pedagogy as the Learning Science Design Review tool: stating this IS the anchor.
0 / 1200 · minimum 20
What your platform can actually observe or log today — not what you wish it could.
0 / 1500 · minimum 20
What you hope the system could adjust on its own — this tool will pressure-test it, not rubber-stamp it.
0 / 600
0 of 3 generations used this session (1 initial + 2 re-runs)
What this designs, and what it doesn't
This designs the adaptation loop — the signals, rules, and review rhythm. It does not run it, and it cannot certify that automated adaptation is appropriate for your context.