What the AI is allowed to do, how to read a decision, and how to agree, disagree or escalate.
15 articles
Sentinel-Q decisions
The exact ceiling on every action Sentinel-Q can take, which ones are never automated at any setting, and why isolating a machine is never done without a person.
Sentinel-Q decisions
Type a sentence, review the exact change and who it affects, then approve it. What Sentinel-Q can draft today, why a proposal goes stale, and how to revise one.
Sentinel-Q decisions
The three settings that decide whether Sentinel-Q tells you, asks you, or acts — where to set them, how to choose, and why only one rule in the product can act on its own.
Sentinel-Q decisions
Every field on a decision, what the confidence figure does and does not mean, and the two or three things to check before you trust one.
Sentinel-Q decisions
What the confidence figure on a decision is worth, what calibration actually means, and why a high number never buys a higher autonomy tier.
Sentinel-Q decisions
How every claim in a decision is checked against your own data, what happens to the ones that fail, and why confidence alone never authorises an action.
Sentinel-Q decisions
The three things you can record on a Sentinel-Q decision, what each one changes, why a reason is compulsory, and why disagreeing is the most valuable button on the screen.
Sentinel-Q decisions
The twelve reasons Sentinel-Q hands an alert to you, in the order it checks them, each with the action worth taking in the next ten minutes.
Sentinel-Q decisions
What a prepared plan contains, how approving and executing differ, what gets audited, and which steps can never be automated at any setting.
Sentinel-Q decisions
The assistant at /ai answers from general security knowledge and never reads your data. What it stores, what it forgets, and where questions about your own estate belong.
Sentinel-Q decisions
Investigations are not billable. What the daily AI budget is for, what a degraded analysis looks like, and which part of the metering is not yet built.
Sentinel-Q decisions
Why an answer got shorter, why the assistant stopped answering, the two per-plan limits behind both, the daily spend caps by plan, and where to watch your spend.
Sentinel-Q decisions
Why an operator typing "eu" into a box does not make inference EU-resident: the four provenance levels, what counts as evidence, and the endpoints that are refused.
Sentinel-Q decisions
Run Sentinel-Q on your own Anthropic, OpenAI, Azure OpenAI or Google Gemini account: the exact fields, why Bedrock is refused, and what happens when your key stops working.
Sentinel-Q decisions
Drive the platform from Claude Desktop or any MCP client: the endpoint, the API key scope it insists on, all fourteen tools, and the twelve capabilities deliberately withheld.