PRESERVEOrigins · March 2024
From coursework to a developing platform.
An account of the original Preserve roadmap, the feedback it received, and how the design is evolving.
The original question
How could AI help people act earlier?
Karl Ohlemann used Preserve, then described in his assignments as Preserve.Inc, as the organization for six assignments in MIT Sloan Executive Education’s Artificial Intelligence: Implications for Business Strategy program. His completion certificate is dated March 2024.
The work explored recurring organizational problems, AI-assisted onboarding, comparative research, and human coaching. It was an academic strategy and implementation roadmap, not an evaluation of a deployed dispute-resolution service.
The 2024 design
AI organizes. People interpret.
Module 1 connected onboarding with identifying recurring problems and equipping human coaches. Module 3 explored research across publications and case law, using church financial mismanagement as an example. Module 5 assigned research and initial analysis to AI while keeping relevance, communication, and interpretation with people.
The final roadmap proposed approximately two to three months of testing alongside human operators once the technology and source material were ready. Early use would assist staff before selected client-facing functions were introduced.
The coursework also addressed role-based access, access logging, and attribution of source material. Broader exercises explored financial health, healthcare collaboration, and agriculture.
Facilitator feedback
A focused, actionable roadmap.
Brian Charles, Head Learning Facilitator, praised the proposal’s clarity, manageable scope, stakeholder benefits, professional organization, and actionability. He recommended expanding the phased implementation timeline.
This was feedback on coursework. It does not establish product performance or imply MIT endorsement, certification, or operation of Preserve.
Current design direction · October 2026
What carries forward—and what changes.
The central idea remains structured preparation supported by AI and interpreted by people. The current platform vision makes adaptive dialogue, exploration of comparable situations, and specialist matching more explicit. The global provider marketplace and commercial model were not detailed in the six assignments.
The revised design would treat examples as prompts for clarification rather than predictions. It would use defined professional-review and escalation requirements instead of relying only on random monitoring as automation expands. It would also recognize that intake can surface urgent safety concerns even when software is not deciding a dispute.
Next steps
Test the usefulness of the handoff.
A proposed pilot would start with a bounded use case and fictional or appropriately authorized material. Practitioners would evaluate factual fidelity, missing information, unsuitable suggestions, escalation decisions, and whether the resulting briefing improves preparation.
Any measured time savings would need to include the work of reviewing and correcting AI output. No pilot results or validated performance metrics are claimed here.