Research Governed

How Adoptic's evaluation framework is built on research-validated variables and auditable evidence.

1. Research-Validated Variables

Adoptic's evaluation framework is built on 73 research-validated variables that have been identified as either causing failure or being necessary for successful adoption of innovations.

Each variable is evaluated across 6 discrete evidence levels, resulting in a total of 365 distinct evaluation points across the full framework.

2. Evidence-Based Scoring

Each variable is scored on a scale from 0 to 5:

Score	Meaning
0	No evidence found in the submitted documents
1	Minimal evidence — early acknowledgement only
2	Partial evidence — some aspects addressed
3	Moderate evidence — substantive but incomplete
4	Strong evidence — well addressed with minor gaps
5	Fully addressed — comprehensive evidence present

3. LLM-Based Pattern Identification

Adoptic uses a large language model (LLM) to identify linguistic patterns within submitted documents. This approach is fundamentally different from generative AI use cases:

The LLM identifies and extracts evidence from the text that has been provided — it does not generate new content or speculate
Because the system is pattern-matching against submitted text rather than generating novel responses, it is not subject to hallucinations
All results are fully auditable — every score is linked back to the specific source text from which the evidence was identified

4. Data Privacy and LLM Usage

Adoptic takes a strict approach to data handling in its use of LLM technology:

Nothing retained by the LLM — no submitted data is stored, retained, or persisted by the LLM provider after processing
No training use — submitted data is never used to train or improve the LLM or any other model
All LLM processing occurs within secured infrastructure with full encryption in transit and at rest

Navigate Back to One Pagers ←