90-Day Pilot
Start with one workflow.
One institution, one champion, one sensitive workflow, one bounded dataset, one measurable outcome — a controlled 90-day decision process.
90-day pilot
A controlled decision process, not an AI transformation program.
Institutional AI adoption should begin with one sensitive workflow and one measurable result.
One institution
One champion
One sensitive workflow
One bounded dataset
One measurable outcome
Days 1–30
Discover + Secure
- Workflow baseline
- Data boundaries
- Security requirements
- Deployment architecture
- Success metrics
Days 31–60
Deploy + Integrate
- Controlled environment
- Institutional connectors
- Model configuration
- Permissions
- Governance
- Initial workflow
Days 61–90
Validate + Expand
- User testing
- Outcome measurement
- Security review
- Deployment assessment
- Production roadmap
Pilot outcome
After 90 days the institution should hold evidence to decide whether the workflow should:
StopIterateMove to productionExpand
Pilot metrics
Measure what changes.
Objectives are defined with the institution before deployment. These are candidate measures, not performance claims.
01
Time saved
02
Cost per workflow
03
Quality improvement
04
User adoption
05
Tasks completed
06
Reduction in manual search
07
Decision-support value
08
Model reliability
09
Security incidents
10
Human override behavior