Mastascusa Holdings · A free tool
Is your business ready for AI?
Answer a few questions about your tools, data, and how work gets done. In about three minutes you get back where you're strong, where you're weak, and what to fix first.
It's an automated estimate, not a review of your actual systems. For that, book a one-hour AI Quick Check.
Scan output
Three production AI surfaces; governance lags behind capability on every pillar.
data architecture
Lvl 2 / 4
Training-serving skew is undocumented for the fraud model; physician-note pipeline lacks formal lineage.
Top gapDefine a baseline distribution per model + monitor in production.
access control
Lvl 2 / 4
Customer-support chatbot can read PHI but role boundaries are inherited from a generic SaaS RBAC.
Top gapEnumerate model-specific access surfaces (weights, fine-tunes, training data, inference endpoints).
process documentation
Lvl 1 / 4
No incident runbook for "the chatbot said something wrong." No kill switch with under-five-minute SLA.
Top gapAuthor and rehearse a model-incident response runbook. Air Canada is a precedent now.
agent governance
Lvl 1 / 4
Three agent surfaces, no published org chart, no named owner per agent, no escalation policy.
Top gapPublish an internal agent org chart with per-agent owner, eval cadence, and escalation path.
Framework crosswalk
| Framework | Status | Note |
|---|---|---|
| NIST AI RMF | partial | GOVERN function mostly absent; MEASURE function ad hoc. |
| ISO/IEC 42001 | gap | No AIMS in place. Pre-certification gap audit recommended. |
| SR 11-7 | gap | Effective challenge requirement not met for any of the three models. |
| EU AI Act | partial | Underwriting summaries may fall under high-risk if used for EU residents. |
| OWASP LLM Top 10 | partial | No documented prompt-injection or supply-chain controls on the chatbot. |
Recommended next step
Book an AI Quick Check: an hour on your tools and data will turn this estimate into a short, concrete plan for a first project.
The Readiness Scan is an automated estimate generated from your inputs. Findings are illustrative and cannot be relied upon for board, regulatory, or commercial decisions. Book an AI Quick Check or a Readiness Review for conclusions based on your actual systems and data.
What happens next
The scan runs and you get a scored topology immediately.
A copy is emailed to your address (and to us).
If the findings warrant a real audit, we follow up.