Reads after-hours emails
Sentinel monitors agreed support inboxes or emergency channels outside normal working hours.
Sentinel AI Agent
After-hours emergency support, without keeping your whole team online
Not every after-hours message is a real emergency. But when it is, your business needs to know quickly.
Sentinel is Centigen's AI emergency response agent. It monitors urgent after-hours emails, checks each issue against escalation rules you approve, and phones your on-call engineer or support contact by voice.
It explains who the message is from, what has happened, why it matters, and what action may be required, and it logs every decision it makes.
Book agent demo
How it worksSee the agent running. Build scope and pricing come from the assessment.
Out-of-hours coverage
Built for teams that cannot afford to miss urgent issues
Many businesses rely on shared inboxes, manual escalation, or employees checking emails outside working hours.
That creates two problems.
First, genuine emergencies can be missed. Second, your team wastes time reacting to messages that are not truly urgent.
Sentinel helps you separate real emergencies from routine requests, then escalates only the issues that need human attention.
It is especially useful for technical support teams, maintenance providers, infrastructure teams, managed service providers, SaaS companies, property management teams, and businesses that need out-of-hours support coverage.
Capabilities
What Sentinel does
Identifies real emergencies
It reviews the message, sender, context, urgency, keywords, business rules, and escalation criteria to decide whether the issue should be treated as urgent.
Filters routine requests
Non-urgent messages can be logged, summarised, and left for the support team to handle during normal working hours.
Phones your on-call engineer
When Sentinel identifies a real emergency, it can place a voice call to your on-call engineer, support manager, or escalation contact.
Explains the issue by voice
The agent tells the engineer who the message is from, what the issue is, what system, asset, site, or customer is affected, and why it requires attention.
Sends a written summary
Sentinel can also send an email, SMS, Slack, Teams, or CRM/helpdesk notification with the issue summary and recommended next action.
Records the escalation
Each escalation can be logged with timestamp, sender, issue type, urgency level, decision reason, and action taken.
Ideal use cases
Where Sentinel earns its keep
Tech support teams
Ideal for support teams that need out-of-hours coverage without running a full night shift. It monitors urgent emails, identifies critical issues, and phones the on-call engineer only when the situation meets your escalation rules.
SaaS companies
Escalate urgent platform issues, critical client problems, service-impacting bugs, payment failures, or operational incidents.
Managed service providers
Monitor client support emails after hours and escalate critical outages, access issues, infrastructure problems, or security-related alerts.
Property and facilities management
Identify emergency maintenance requests, water leaks, access issues, power problems, security concerns, or health and safety risks.
Engineering and maintenance teams
Route urgent equipment, site, or asset issues to the right on-call engineer.
Logistics and operations teams
Escalate urgent operational failures, shipment disruptions, warehouse issues, or customer-critical incidents.
Decisioning
How Sentinel decides what is urgent
Sentinel is configured around your business rules.
It can assess:
- Sender type
- Customer priority
- Keywords and issue category
- Severity language
- Affected system or location
- Business impact
- Time sensitivity
- Contract or SLA rules
- Known emergency scenarios
- Escalation thresholds
For example, a routine password reset may be logged for the morning. A system outage affecting a key client may trigger an immediate voice call to the on-call engineer.
Why it matters
A structured way to handle urgent issues
After-hours support is difficult to manage. Full coverage is expensive, but relying on people to manually check emails is unreliable.
Sentinel gives your business a structured way to manage urgent issues outside office hours.
It helps you:
- Reduce missed emergencies
- Avoid unnecessary wake-ups
- Improve response time
- Protect key clients and assets
- Reduce pressure on support teams
- Create a clear escalation trail
- Improve out-of-hours service coverage
- Keep routine requests for normal working hours
How it works
From escalation rules to live cover in six steps
1
We define your escalation rules
We map what counts as an emergency, what should be ignored until morning, and when Sentinel should call someone.
2
We connect your support inbox
Sentinel is connected to the agreed email inbox or support channel, with clear rules for after-hours monitoring.
3
We configure your on-call contacts
We add your escalation contacts, on-call engineers, backup contacts, and routing rules.
4
We train Sentinel on your business
We add your support categories, customer priority rules, SLA logic, common emergency scenarios, and approved escalation language.
5
We test real scenarios
We test urgent, borderline, and non-urgent messages to make sure Sentinel escalates correctly.
6
You go live
Once approved, Sentinel monitors after-hours emails, filters routine messages, and phones the right person when a real emergency is detected.
Why governance
What goes wrong when out-of-hours triage is ungoverned
An agent that reads your support inbox overnight and decides who gets woken is making operational decisions in your name. Run without scoped authority, named contacts, and a written record, those decisions still happen, they just happen unexamined. These are the failure modes that show up most often.
Missed escalations
A genuine emergency arrives at 2am, gets misjudged as routine, and sits unread until morning. Without approved, tested escalation thresholds, nobody knows the miss happened until the customer reports the damage.
Unlogged decisions
An issue is downgraded overnight and left for the day team. If the decision is not recorded with its reason, nobody can say afterwards what was assessed, which rule applied, or whether the call was right.
Escalations to the wrong person
Calls go to whoever was configured months ago rather than tonight's on-call engineer, with no backup chain when there is no answer. The escalation technically happened; the response never did.
Silent drift
Customers, SLAs, and systems change, but unowned triage rules quietly do not. The agent keeps making confident decisions against last quarter's priorities, and the gap only surfaces in an incident review.
The controls
Sentinel runs under controls you set
Sentinel's authority is scoped before it goes live. It monitors the inboxes you name, applies the escalation rules you approve, and takes no action outside them.
Nothing is approved at 2am, because everything is approved in advance. You sign off the emergency scenarios, the escalation thresholds, and the approved language before Sentinel goes live, and any change to those rules comes back to you. Sentinel never asks for permission out of hours, it acts only inside permission you have already given.
Escalation always lands with a named human. Sentinel calls the engineer on your on-call rota, follows your backup chain if there is no answer, and your team keeps control of the final response.
And every decision, escalated or not, is written to the audit trail, so the next morning you can see exactly what was assessed, what was done, and why.
Escalation to named humans
What your on-call engineer receives
When Sentinel calls your engineer, it can explain:
- Who sent the message
- Customer or site name
- Contact details
- What the problem is
- When the message was received
- Why it was treated as urgent
- Affected system, service, location, or asset
- Recommended next action
- Whether any backup contact should be notified
The engineer does not need to search through the inbox first. They receive the critical context immediately.
Audit trail
Written summary and escalation log
After an issue is assessed, Sentinel can create a written record of the decision.
Each log can include:
- Sender
- Subject
- Message summary
- Urgency level
- Decision reason
- Escalation status
- Person contacted
- Call outcome
- Time of escalation
- Recommended next step
This gives your team visibility and accountability for after-hours decisions.
Running costs
The governed agent retainer
Governed agent retainer
$1,250/month
The retainer covers:
- Monitoring of Sentinel's decisions and performance
- Escalation handling and on-call rota changes
- The full audit trail, maintained and reviewable
- Approved knowledge base maintenance: escalation rules, scenarios, and priorities
The build itself (inbox connection, escalation rule mapping, on-call contact setup, voice call configuration, and pre-launch testing) is scoped and quoted through the Agent Readiness Assessment.
FAQ
Frequently asked questions
Sentinel is designed for urgent after-hours triage. It can read support emails, decide whether the issue meets your escalation rules, and call the right person when action is needed.
Yes. Sentinel is a strong fit for tech support teams that want out-of-hours coverage without staffing a full support desk overnight.
Yes. When an issue meets the agreed escalation rules, Sentinel can phone your on-call engineer or support contact and explain the issue by voice.
It can explain who the message is from, what the problem is, why it was treated as urgent, and what action may be required.
Yes. Routine or low-priority messages can be logged and left for the team to handle during normal working hours.
Yes. Sentinel is configured using your escalation rules, customer priority levels, SLA logic, support categories, and approved scenarios.
Yes. Sentinel can be connected to your support workflow depending on your tools and integration requirements. This may include helpdesk, CRM, email, Slack, Teams, or other systems.
Yes. Sentinel can follow agreed escalation rules, including primary and backup contacts.
The Agent Readiness Assessment takes 10 working days and produces the scope and quote for your build. Most deployments can then be prepared in around two weeks, depending on complexity, inbox access, escalation rules, and voice call configuration.
The retainer is $1,250 per month. It covers monitoring of Sentinel's decisions, escalation handling, the full audit trail, and maintenance of the approved knowledge base: your escalation rules, scenarios, and priorities. All pricing is in USD.
Yes. We test urgent, non-urgent, and borderline scenarios before launch.
See it triage a 2am emergency.
Book a demo and we will walk you through Sentinel reading an after-hours message, deciding whether it is a real emergency, phoning the engineer on call, and logging every step.
Book agent demo
See the agent running. Build scope and pricing come from the assessment.