AI

ArkChat: ask Ark about your infrastructure, see how it answered

Questions answered from your live platform data, not from general knowledge, with the query path shown so the answer can be checked rather than trusted.

Short answer

What is ArkChat?

ArkChat is the conversational interface to DevOpsArk that answers questions about your own infrastructure (clusters, services, deployments, incidents, cost and security posture) using live platform data, and shows the steps it took to produce each answer.

Why it matters

What ArkChat is for

The conditions this module removes. If none of these are familiar, you probably do not need it yet.

  • Answering "which services are affected by this CVE" means opening four consoles and joining the results by hand.
  • Only two people know the query language well enough to get answers quickly.
  • General-purpose assistants know Kubernetes but know nothing about your clusters.
  • During an incident, the person who can find the data is the bottleneck.
  • An answer from a model that cannot show its working is not usable for an operational decision.
How it works

ArkChat resolves a question into queries against the DevOpsArk data model (the application catalogue, cluster inventory, metric and log stores, deployment history, cost allocation and security findings), and answers from the results. It does not answer infrastructure questions from general model knowledge, because general knowledge does not contain your cluster. Every answer is accompanied by the execution trace: which queries ran, against which sources, and what they returned, so the answer can be verified rather than trusted. Where an action is possible, ArkChat proposes it as a plan with an explicit scope and asks for approval; it does not execute a change because a sentence sounded like a request. Access follows the asking user own permissions, so ArkChat cannot reveal data the person could not otherwise see.

Capabilities

What ArkChat does

The 7 capabilities that make up ArkChat.

Answers from your data

Questions resolve to queries against your live inventory, metrics, logs, deployments, cost and findings, not to general knowledge.

Visible execution trace

Every answer shows the queries that produced it and what they returned, so it can be checked.

Proposed plans, not silent actions

Where an action is possible it is proposed with explicit scope and requires approval before anything runs.

Permission-scoped

ArkChat answers within the asking user own access. It cannot surface data the person is not entitled to see.

Incident companion

During an incident, ask for the recent deployments, the correlated alerts and the matching log patterns in one place.

Works where you work

Available in the DevOpsArk console and in the chat tools your team already uses.

Audited conversations

Questions, the queries they generated and any approved actions are recorded.

Architecture

How ArkChat fits together

Question
ConsoleSlackMicrosoft Teams
ArkChat
Intent resolutionQuery plannerPermission filterTrace recorder
Data sources
Application catalogueCluster inventoryMetrics and logsDeploymentsCostFindings
Action
Proposed planApprovalAudited execution
ArkChat architecture within the DevOpsArk control plane.

Outcomes

  • Infrastructure questions are answerable by anyone on the team, not only the two people who know the query syntax.
  • Answers can be verified, which is what makes them usable operationally.
  • Incident data gathering stops being a bottleneck.
  • Nothing changes without an explicit approval.
  • Access control is preserved rather than bypassed by a convenience layer.
How to use it

Using ArkChat, step by step

The path from connecting a source to getting value, in the order it happens.

  1. 1
    Ask

    Put the question in plain language from the console or a chat tool.

  2. 2
    Plan the query

    ArkChat resolves the question into queries against the platform data model.

  3. 3
    Filter by permission

    Results are limited to what the asking user is entitled to see.

  4. 4
    Answer with a trace

    The answer is returned alongside the queries and results that produced it.

  5. 5
    Propose an action

    Where a change is possible, it is proposed with scope and requires approval.

Use cases

Where teams apply ArkChat

On-call

Gather incident context fast

Ask what deployed in the last hour, which alerts correlate and what the logs show, in one exchange.

Security

Scope an advisory

Ask which services ship an affected library and which of them are internet-reachable.

Engineering leadership

Get an answer without a ticket

Ask what a team spend was last month, or how many services are missing resource limits.

New joiner

Learn the estate

Ask what a service depends on and where it runs, and get an answer drawn from the live system.

Supported technologies

What ArkChat works with

Named integrations link to their own page. The rest are supported runtimes and formats.

Do not see your stack? DevOpsArk works over standard interfaces: the Kubernetes API, OCI images, OpenTelemetry and cloud provider APIs, so most environments are supported without a bespoke connector. Ask us about yours.
FAQ

ArkChat: frequently asked questions

The 8 questions teams ask most often before adopting ArkChat.

See ArkChat against your own environment

A 30-minute walkthrough with a platform engineer, not a sales deck. Bring a cluster and a problem.