OPEN RESEARCH PROTOCOLS

Evidence that can
be challenged.

Causivra separates documentation, direct observation and editorial judgment. These protocols define what we measure before a result exists, so a conclusion cannot quietly rewrite the test.

STANDARDSame scenario for every candidate
DISCLOSUREAssistance and limitations recorded
RESULTSNo publication without a usable sample
01

Software adoption test

Can real users complete the work without expert assistance?

  • Task completion rate
  • Median completion time
  • Help requests
  • Critical errors
  • User confidence after the task
02

Automation reliability test

Does the workflow remain safe when data or dependencies fail?

  • Successful runs
  • Silent failures
  • Manual recovery time
  • Duplicate actions
  • Time to detect an exception
03

24-month cost model

What will the operating model cost after the trial ends?

  • Licences and usage
  • Implementation labour
  • Administration
  • Integration and support
  • Migration and exit cost
04

AI output review

How much verified work survives human review?

  • Supported factual claims
  • Material omissions
  • Correction time
  • Unsafe actions blocked
  • Accepted output without revision
PUBLICATION RULE

What counts as evidence

Documented fact links to a first-party or authoritative source and carries a review date.

Observed result identifies the scenario, sample, environment and material limitation.

Editorial judgment names the buyer context and remains separate from commercial compensation.

Unresolved claim is excluded or clearly marked—not converted into a confident conclusion.

CONTRIBUTE A REAL CASE

Bring a workflow.
Keep the test honest.

Organizations and vendors may suggest a documented scenario or correction. Participation never guarantees a positive result or ranking.

Propose a test or correction ↗