How Sentra raises problems before anyone asks
Every AI tool waits for a question. The costly problems in a company are the ones nobody knows to ask about. Sentra looks for them, checks they are still open and tells the person who needs to know, and stays silent when there is nothing worth saying.
In short
- On a schedule each person sets, by default a weekday morning, Sentra picks the one thing that person most needs to know today from their calendar and the company's memory.
- Before it says anything, it re-checks that the item is still open and drops it if it may already be closed. If nothing qualifies, it sends nothing.
- For each project, Sentra looks for risks and contradictions in what the viewer is allowed to see. A finding is kept only if two of three independent checks agree, and each one stores its evidence.
- These checks are read-only. They cannot send email or Slack messages, edit documents or change tasks.
Why the value is in what nobody asked
The companies that get the most from Sentra found that the value was not in faster answers to questions they already had. It was in what Sentra raised that nobody had asked about: a problem that sat between two teams or two systems, which each side saw only part of.
In an independent test built and scored by Era by Eon, Sentra was connected to a simulated 1,200-person company it had never seen, asked nothing, and found 495 of the 497 planted problems. Claude Opus with the same access and no Sentra found 8%. On direct questions the two tie. The difference is everything nobody thought to ask.
The daily message
Each person chooses when Sentra checks in. The default is weekdays between 6 and 8 in the morning, local time. Sentra reads that person's calendar and the parts of company memory they may see, and picks the one item that most needs their attention today.
- It re-checks that the item is still open and drops it if it may be resolved
- It keeps the message to one or two lines
- If nothing qualifies, it stays silent rather than sending filler
- Messages are at least an hour apart, and Sentra pauses after two go unanswered
- It arrives where the person already is, such as iMessage or WhatsApp
Risks and contradictions on every project
Separately, Sentra watches each project or workstream for risks and for contradictions, such as two teams working from different dates for the same launch. Each finding must be agreed by two of three independent checks before it is kept, a contradiction must name the decision that is at stake, and every finding stores the evidence it rests on and a confidence level.
A person stays in control
Finding a problem and acting on it are kept apart. Proactive checks are read-only: they cannot send email or chat messages, write documents or change tasks. When Sentra's agent does act, such as drafting a reply, a named person approves it before it goes anywhere.
Frequently asked questions
What does proactive AI mean in Sentra?
Sentra checks the company's memory for what needs attention without being asked, re-checks that each item is still open, and tells the person who needs to know. It stays silent when nothing qualifies.
Can proactive checks send messages or change things on their own?
No. They are read-only and cannot send email or chat messages, edit documents or change tasks. Anything Sentra's agent sends is approved by a named person first.
Will it flood people with alerts?
No. The daily message is one or two lines about the single most important item, messages are at least an hour apart, and Sentra pauses after two go unanswered.
How is this different from scheduling a chatbot every morning?
A scheduled chatbot starts from scratch each time and only finds what its prompt tells it to look for. In Era by Eon's test, Claude Opus given a checklist found 24% of the planted problems. Sentra works from a memory that already connects people, accounts and systems, and found 495 of 497.