When to use the pipeline journey

The pipeline journey is CXGear’s step-by-step timeline for one record. It shows each step’s result, messages, and timing. It is powerful — and it is also easy to overuse.

This page helps you decide when to open it, and when the Records detail panel is enough.

What the journey is

For a single contact in a single job, the journey shows:

Step name and outcome (for example log · logged, ai_agent · ai_completed)
Message text (log templates, function console lines, AI errors)
Started / ended / duration for the whole record
Final output object used for debugging

It is a debug and build tool, not the main operations screen.

Screenshot needed

Journey drawer showing steps with messages and time taken.

Where: Records detail → Pipeline journey

Save as: src/assets/screenshots/17-journey-drawer.png

When you should open the journey

Open Pipeline journey when you need to answer questions like:

Situation	Why journey helps
Record failed	See which step failed and the error text
Outcome is blank or surprising	Check whether a Tag, AI, or channel step actually ran
You are building or testing a spinner	Confirm log templates and function output
AI step misbehaved	Read provider errors (wrong model, missing key, timeout)
Timing looks wrong	Compare step durations and overall record time
Support is investigating a ticket	Share the exact step messages, not just the outcome column

If you are actively changing a spinner, journey is your best friend. Log steps and function console lines show up here clearly.

When you should not open the journey

Stay on Records (list + detail panel) when you only need:

Who was contacted (metadata)
What happened operationally (disposition, AI reply, message id)
When it finished
Which spinner ran

Supervisors reviewing hundreds of contacts should not open a journey for every row. That is slow and noisy. The detail panel is designed for that scan-and-review loop.

See Opening record detail.

How to open it

From Records:

Click a row to open the detail panel below the table.
Click Pipeline journey.

From Jobs:

Select a run on the left.
Open a record on the right.
Open the journey from that record’s actions.

Either path leads to the same per-record timeline.

Success looks like

Ops reviews use Records detail by default
Journey opens only for failures, tests, and unclear outcomes
Builders use log steps so journey messages are readable
You can explain a failure from journey text without guessing

Common problems

Problem	Likely cause	What to do
Journey is empty or thin	Steps did not emit messages	Add a Log step while testing
AI error is cryptic	Provider rejected the model or key	Fix the LLM integration; retry
Opening journey for every row	Treating journey as the ops UI	Use detail panel for outcomes
Cannot find journey	Clicking the row only	Open detail first, then Pipeline journey
Old run still fails after a fix	Finished jobs are historical	Publish / run again and inspect the new record

Journey vs Jobs vs Records

Screen	Best for
Records	Daily ops: who / what / when
Jobs	Run health: progress, duration, batch status
Pipeline journey	One-record debug: step messages and errors