0 / 20 answered
01 / Context Integrity
How well shared context is maintained and carried across sessions.
I provide the AI with a structured summary of prior decisions and context at the start of each new session.
AI outputs remain consistent with directions and constraints set in earlier sessions.
I have noticed the AI producing outputs that contradict or ignore decisions already established in the collaboration.
There is a documented record of key decisions from AI sessions that I can refer back to at any time.
02 / Memory Governance
Whether corrections and decisions are preserved and accessible across sessions.
Corrections I make to the AI in one session are applied and available in future sessions.
I maintain a persistent external record of important working decisions — separate from the AI's context window.
If all AI context were lost today, I could reconstruct the full working state of my project from existing records.
I have a defined process for updating shared working memory when decisions change or new constraints are established.
03 / Verification
Whether AI outputs are independently checked before being acted on.
I treat verification of AI outputs as a distinct step — separate from the generation process.
Numerical or factual claims in AI outputs are checked against source material before I act on them.
I rely on the same AI that generated an output to also verify that output.
Critical decisions based on AI outputs are reviewed independently before action is taken.
04 / Language Governance
Whether instructions are precise and shared vocabulary remains stable across sessions.
The AI and I use defined, consistent terminology across all sessions — terms are not left to be interpreted each time.
My instructions to the AI are specific and unambiguous — I avoid relying on the AI to infer intent.
I correct imprecise, vague, or drifting language when it appears in AI outputs — rather than accepting it and continuing.
I have noticed the same terms or concepts being interpreted differently by the AI across different sessions.
05 / Recovery
Whether you have defined procedures for when collaboration failures occur.
I have a defined procedure for recovering from a failed or corrupted AI collaboration session.
When something goes wrong in an AI collaboration, I can identify what type of failure has occurred.
I have defined checkpoints in my AI workflow that allow me to restore a known-good working state.
A major AI collaboration failure has caused me to lose significant work or restart a substantial piece of work from scratch.
Answer all 20 questions to continue.
Your Results
/ 100
Score by domain
Failure modes you are currently exposed to
Recommended reading for your gaps

Based on your lowest-scoring domains, these papers address the structural problems your workflow is currently exposed to.