Miragon AILifecycle
OptimizeLearn from what runs.
In Optimize your team learns from the processes in production what to change in them, and the loop starts again in Design. Miragon has no tool for this phase yet.
The problem today
There is no fixed way back from operations to the model
Cockpit shows you, one engine at a time, what is running and what is failing. Questions such as "Where do instances spend their time?" or "Did the last deployment make things worse?" usually need someone who writes PromQL against Prometheus or queries the engine's history API. When a number stands out, someone carries it by hand to the team that changes the model.
No tool yet
Miragon has no tool for Optimize today, and we say so on purpose. The lifecycle map on the landing page leaves this slot open. The analytics module in Pulse stays in Execute, because calling it the Optimize answer would promise more than it delivers.
What this phase needs
The questions an Optimize tool would have to answer
Each role brings its own question to operations.
- Product ownerDoes the new version run faster, or was it the time window?
- Automation leadWhere does a case lose time, in the work itself or while waiting in front of a user task?
- ArchitectWhich paths and variants does the process take in production?
- Lead developerWhat effect did the changed DMN rule have?
- Platform leadWhich metrics do we keep, and for how long, so that comparisons span months?
- Head of ITWhich change pays off first, and how do we see that it worked?
What Execute already hands over
Raw material for some of these questions
- Pulse, in ExecuteThe analytics moduleRanks activities by execution time and incident rate, draws a heatmap on the BPMN diagram, and compares versions, time windows and engines, or the figures before and after a deployment.More on Pulse
- operate, in ExecuteHistory queriesRuns history queries such as operate historic-process-instance list and returns the raw results as JSON, without KPIs.More on operate
The analytics module needs a CIB seven engine with the metrics plugin io.miragon.mcp:cibseven-history-metrics, which exports through Micrometer to Prometheus. The plugin exists for CIB seven only. The module looks back 30 days at most, and data from before you installed the plugin is missing. It shows no waiting times, paths or variants, no business data such as product or channel, and no run times per activity and version. Both Execute tools are pre-1.0 (Pulse v0.20.0, operate 0.7.0).
Human and AI
Your team weighs the findings
- An agentCould turn your questions into queries and lay the numbers side by side without remodelling anything itself, but no tool for that exists yet.
- Your teamWeighs the findings and decides what goes back into Design.
Hand-over
From raw figures to a change in Design
Comes in
The raw material described above: Pulse's activity ranking, BPMN heatmap and KPI deltas, and the raw results of operate's history queries.
- JSON and chat widgets (Pulse)
- JSON (operate)
Bring us a question from your operations.
We decide from real cases what to build for Optimize, if anything. If your team faces one of these questions right now, we look at your process and your data with you and work out which question needs an answer first.
Talk to usA 30-minute first call.