LoginSign Up
← Back to all sparks
A

Axiom

ANALYTICS
Velocity6.3

Log management and observability platform for events and traces.

axiom.co ↗

Axiom is rebuilding observability so an AI agent, not a human, can be the first user.

observabilityagent-nativemcpdashboardsmetricsevaluations
◆Current state
Axiom is a logs, traces and metrics platform that reached feature parity on the fundamentals earlier this year — metrics went generally available in March, dashboards got a full API, and Correlations tied the three data types together for investigations. The last two months have been spent thickening the console: collapsible dashboard sections, gauge elements, schema locking, Grafana as a query surface. Underneath that steady product work, a second track has been running the whole time, aimed at AI agents as operators rather than at humans.
◆Where it's heading
That second track is now the main story. Metrics shipped queryable by agents through MCP and a dedicated skill, monitor management moved into the agent surface alongside the Grafana work, and evaluations arrived as both a live-traffic scoring feature and an agent-authored skill. The August release takes it to the account layer: an agent can now create its own Axiom organization and have a human claim it afterwards. Axiom is systematically removing the assumption that a person is present at each step.
◆Prediction
The remaining human-gated surfaces are billing, access control, and dataset provisioning, and agent-created orgs makes those the obvious next targets. Expect the skills catalogue to keep growing into a set of task-shaped agent entry points rather than a single MCP endpoint.

◆Recent moves

  1. 12d ago

    Agent-created orgs, richer charts, sharper queries

    ⚡ SPARK

    The release bundles chart legend controls, trace zoom and an `in` filter for metrics, but the item that matters is agent-created organizations. It moves the agent-native track from querying and monitoring, where it has lived all year, up to account provisioning itself.

    View source ↗
  2. 1mo ago

    Dashboard sections

    Dashboards can be split into collapsible sections that stop running their queries while closed. It is a usability change with a real cost effect on large dashboards, and part of the console-thickening work that ran alongside the agent track.

    View source ↗
  3. 1mo ago

    Gauge dashboard elements

    Adds a gauge element with configurable segments, labels, colors and an optional sparkline. A straightforward gap-fill in the dashboard component set rather than a change in what Axiom can measure.

    View source ↗
  4. 1mo ago

    Dataset schema locking

    Datasets can be pinned to a declared schema so their shape stays predictable, shipped with a lighter console, smarter query editing and separate view and edit modes for dashboards. Schema locking matters most for the machine consumers Axiom has been courting, since a stable shape is what makes generated queries safe to run.

    View source ↗
  5. 1mo ago

    APL and MPL in the Grafana data source

    The Grafana data source gained both APL and MPL querying, plus four weighted aggregations in APL and monitor management from an AI agent. The Grafana half broadens where Axiom data can be read; the monitor-management half is an early step on the same agent-operator path the August release extends.

    View source ↗
  6. 2mo ago

    Correlations

    Correlations link logs, traces and metrics datasets so an investigation can move from a first clue to surrounding system state without hand-written query stitching. It closes the gap between the three data types Axiom had been storing separately and is the groundwork any automated investigation would need.

    View source ↗