What is an incident context lake?
An incident context lake is one place where the signals around an incident come together, and Context Lake is Noci AI's own name for it. Agents and people start an incident from the same facts, and what happened and what was done is recorded in that same place.
What does an incident context lake hold?
At the start of an incident the facts are scattered. The alert is in a monitoring tool, the deploy in a pipeline, the customer’s complaint in a ticket queue, the conversation in chat, and the owner of the service on a roster. An incident context lake brings them into one place, so nobody has to collect them before they can start. Typically it holds:
| What it holds | Why an incident needs it |
|---|---|
| Alerts | What fired, when, and how often it has fired before |
| Deploys and code changes | What changed shortly before the alert |
| Tickets | Whether customers are already reporting a problem |
| Chat and meetings | What people have already tried and said |
| Ownership | Who to ask, and who to page |
In a good one, each fact is tied to the service it concerns, so a question about one service finds its alert, its latest change and its tickets together.
Why start an incident from one set of facts?
Without a shared record, the first minutes go to collecting context: opening several tools, asking in chat who changed what, and repeating the answer for each person who joins. When agents and people read the same facts, the page, the incident channel and the investigation begin with context instead of a question.
The incident also gets a record. What happened and what was done is written back to the same place, rather than living only in a chat thread or on one person’s screen. It also makes a handoff plainer: a person who takes over reads the same record as the person who started, not a retelling of it.
What is Noci AI’s Context Lake?
Context Lake is Noci AI’s own name for its incident context lake. It is the system of record for incidents and alerts, and the outcomes that agents and people produce are written back to it. Noci AI’s agents investigate with your telemetry and your recent code changes, and what they find is written back to the same place.
It is included in the price, and the bands are on the pricing page.
What does it hold today?
- Technical signals. Alerts and telemetry from the tools you connect, and the code changes around them. The integrations page lists every tool.
- Customer tickets. Pulled from the support tools you connect, so they are part of the incident’s context.
Not available yet: Customer impact from billing and CRM Coming.
What should you ask about any incident context lake?
- What feeds it? Know which tools it reads, and whether you can see where each fact came from.
- Are the facts joined? A pile of events is not a context lake. Look for facts tied to the service and its owner.
- What is written back? The record of an incident should include what the agents did, not only what the alert said.
- Who decides what agents do with it? With Noci AI, you choose, step by step, whether an agent asks before it acts. To test an agent before it goes near on-call, see how to evaluate an incident agent.
Questions
Is an incident context lake the same as a data lake?
No. A data lake keeps data to analyze later. An incident context lake is organized around an incident: it joins the alert, the change, the ticket and the owner for a service, so the people and agents working the incident start from the same facts.
What is Noci AI's Context Lake?
It is Noci AI's name for its incident context lake. Alerts and telemetry, code changes and support tickets from the tools you connect come together there, and agents and people write outcomes back to it.
Does the Context Lake use credits?
No. The Context Lake, Noci's map of your services, is included and never uses credits. Credits pay for minutes of AI work, such as an investigation.
Which tools feed it?
The alert and telemetry, code and support-ticket tools Noci AI integrates with. The integrations page lists every tool.
Who decides what agents do with it?
You choose, step by step, whether an agent asks before it acts.