Skip to content

Ask Ellie

Engineering answers from your systems of record.

Ask Ellie answers questions across repositories, trackers, incident tools, observability, and agent sessions, with citations to the source records.

  • GitHub, read for this answer
  • GitLab
  • Linear
  • Jira
  • Sentry, read for this answer
  • Datadog, read for this answer
  • Claude
Ask Ellieconnecting push → panic → 5xx
  • GitHub, read for this answer
  • GitLab
  • Linear
  • Jira
  • Sentry, read for this answer
  • Datadog, read for this answer
  • Claude
Ask Ellieconnecting push → panic → 5xx
Ask Ellie · web appINC-5120

What broke after the 11:05 config push?

The push shipped a 131-entry features file from PR #7312. edge-proxy caps it at 100 and panicked on load in every region. INC-5120 opened at 11:12; the last good file was restored at 11:48.

  • PR #7312
  • Sentry · edge-proxy
  • Datadog · 5xx
Ellie#edge-oncall

11:05 config push → 131-entry file → edge-proxy panics. INC-5120 resolved; postmortem needs an owner.

Board · Incident response
peak 5xx
41%
push → errors
2m
to restore
43m
  • 01

    Cited sources

    Every answer links to its source records.

  • 02

    Dashboard integration

    Open Ask Ellie from a dashboard widget with its data attached.

  • 03

    Report mode

    Generate long-form reports, including postmortem drafts.

  • 04

    Slack integration

    Ask from any Slack channel where Ask Ellie is installed.

World Model

Built on the Entelligence World Model.

The World Model links code, tickets, deploys, incidents, and agent sessions across connected tools. Ask Ellie answers from that graph.

  • code
  • tickets
  • deploys
  • incidents
  • agent sessions

Incidents

  • Summarize open Sev-1 and Sev-2 incidents
  • Which incidents need attention right now?

Delivery

  • Where are we leaking the most time in delivery?
  • Which PRs are at risk of not being merged?

Deployments

  • What broke after the 11:05 config push?
  • Show me the DORA metrics for our team

AI usage

  • How much AI-generated code hit production?
  • Which teams are over capacity?
  • 01

    Incidents

    • Summarize open Sev-1 and Sev-2 incidents
    • Which incidents need attention right now?
  • 02

    Delivery

    • Where are we leaking the most time in delivery?
    • Which PRs are at risk of not being merged?
  • 03

    Deployments

    • What broke after the 11:05 config push?
    • Show me the DORA metrics for our team
  • 04

    AI usage

    • How much AI-generated code hit production?
    • Which teams are over capacity?

Surfaces

Built into the tools your team already uses.

Ask Ellie answers in the web app, in Slack, and in your terminal. Every surface reads the same connected data.

  • 01

    Write it up in the web app

    Every question stays in history, and Report mode turns an answer into a write-up.

Ask Ellie across Entelligence

Each Entelligence product adds its data to the same graph.

How it works

From a question to a cited answer.

Ask Ellie reads your connected tools, links the records it finds, and cites each one in the answer.

Ask from any surface. Ask Ellie a question in Slack, the web app, or CLI. Use natural language, no special syntax.

SlackWeb appCLIWhy did the API slow down this morning?Ask in plain language

Connections

One answer, read across every tool you connect.

Ask Ellie reads only the tools connected to your workspace. Switch one off and its records leave the answer.

Ask Ellie

What happened with edge-proxy this morning?

  • The 11:05 push shipped a 131-entry features file from PR #7312.1
  • edge-proxy panicked on load: 131 entries, limit 100.2
  • Global 5xx went from 0.2% to 41% in four minutes.3
  • INC-5120 is resolved. The size-check follow-up, EDGE-231, is still open.4
Cited 4 of 15 connected tools.

Security

Security and data controls.

  • 01

    Cited sources

    Every answer links to the underlying records.

  • 02

    Scoped access

    Ask Ellie reads only the systems connected to the workspace.

  • 03

    No model training

    Customer code is not used to train models.

  • 04

    SOC 2

    Independently audited. Report available on request.

The questions teams ask before turning on Ask Ellie.

Missed one? Ask it in the demo.

  • Engineering leaders, managers, and the engineers on call. Leaders ask about delivery, DORA metrics, and AI adoption; managers ask which pull requests are at risk and which teams are over capacity; on-call engineers ask what changed before an incident.

  • Only the systems connected to the workspace: code hosts such as GitHub and GitLab, trackers such as Linear and Jira, incident and observability tools such as Sentry and Datadog, AI coding agent sessions, and the data from other Entelligence products. Disabling a connection removes its data from every answer.

  • In the Entelligence web app, in Slack channels where Ask Ellie is installed, and in the terminal through the Ellie CLI, which can also share context with Claude Code, Codex, and Cursor over MCP. Every surface uses the same connected data.

  • No. Ask Ellie opens from a dashboard widget with that widget’s data attached and answers the questions a chart raises, such as which deploys, pull requests, and incidents sit behind a change on the metric timeline.

  • No. Customer code is not used to train models. Ask Ellie reads only the systems connected to the workspace, and Entelligence is independently audited for SOC 2, with the report available on request.

Evaluate Ask Ellie on your own engineering data.

Connect your systems of record. Ask Ellie is available in the web app, Slack, and the CLI.

  • Cited sources
  • Scoped access
  • No model training
  • SOC 2