Future-state alignment artifact
eRx AI-native flow principles
Three black-and-white diagram directions for aligning on the interaction contract before selecting a detailed flow.
Shared contract
Otto prepares relevant context and makes its interpretation visible. The provider retains clinical choice, correction, approval, and signing. Recovery and later work preserve the original decision context.
Option 01
Decision spine
Can we align on the full family as one sequence of meaningful provider decisions?
Recognize
treatment, blocker, or return
Prepare
context, source, freshness
Provider decides
start, correct, approve
Preflight
safety, access, authority
Sign and send
final order and status
Return later
owner, reason, next action
Recovery rail: pause, correct, cancel, replace, or safely retry without duplication.
Best when the team needs one shared end-to-end backbone.
Option 02
Authority loop
Can we align on the handoff between system work and provider authority at every moment?
1. Context arrives
patient, treatment, pharmacy, benefit, safety, authority
2. Otto shows its work
understood, prepared, changed, uncertain, or unavailable
3. Provider directs
confirm, correct, change, defer, stop, or hand off
4. Canvas responds
re-prioritizes context, advances safely, or preserves a return point
The loop repeats at readiness, composition, safety, signing, and recovery.
It makes the boundary unmistakable: Otto can prepare and explain, while the provider remains accountable for consequential action.
Best when the main alignment risk is unclear AI authority.
Option 03
Continuity map
Can we align on how one decision becomes safe asynchronous follow-through rather than an abandoned task?
Now
One active decision
the provider sees only the context and action that matter now
Otto
Preserves an accountable return point
source decision, owner, reason, status, and what already happened
Later
Resolved
confirm outcome and complete
Needs decision
resume in original context
If blocked
Correct
change intent
Redirect
new pharmacy or plan
Recover
cancel or safe retry
Best when the team needs to make returns, ownership, and recovery concrete.
Suggested use in the alignment session
Start with the decision spine to agree on scope. Use the authority loop to test the human and AI boundary. Use the continuity map to test whether every branch retains context and a safe next action. All three are future-state hypotheses, not current product behavior or customer validation.