Runs & review
How to read the run history, run a fifteen-minute daily review in the first week, trace a failure to its cause, and pause an agent when you need to.
On this page
Every piece of work an agent does is recorded as a run. The run history is how you see what booked, what dropped, and what still needs a person.
#Run outcomes
| Outcome | What it tells you | What to do |
|---|---|---|
| Completed | The job finished as the rules describe | Nothing. Spot-check a few each day in week one |
| Refused | The agent correctly said no | Check the refusal was right, not a stale schedule |
| Handed off | A rule sent the thread to a person | Make sure the owner picked it up in time |
| Corrected | Someone on your team fixed the result | Tell us, so we fix the source and not the symptom |
| Abandoned | The customer stopped replying | Look for a pattern: a long question, a slow answer |
#The daily review
For the first week, review runs at the same time each day. Fifteen focused minutes is more useful than a weekly meeting built from memory.
Scan the corrections
Anything your team had to fix comes first. Each one is a source to change.
Check the handoffs
Did every handoff reach its owner, and was the reply on time? Repeated handoffs of the same kind are worth a rule.
Read a few completed runs
Open three or four at random and compare them to your calendar: service, provider, location, time and duration.
Note anything stale
A changed price, a new policy, a provider on leave. Send it to us the same day.
#Tracing a failure
When a run goes wrong, we trace it to the nearest cause and fix that, instead of piling more instructions onto the conversation.
| Cause | Example | Fix |
|---|---|---|
| Source data | A provider's hours are wrong in the calendar | Correct the calendar |
| Service mapping | "Tidy up" mapped to the wrong duration | Update the mapping |
| Policy | Two staff answer the same case differently | Settle it in the brief |
| Tool | A write did not land as confirmed | Fix the integration, re-run the test |
| Handoff ownership | A handoff sat with nobody | Name an owner and a backup |
| Scope | The agent was asked to do a job it should not | Narrow the scope |
#Pausing an agent
Before launch you agree a pause procedure: who can stop new calls, messages or calendar writes, and what customers hear while it is paused.
While an agent is paused, your phone line, POS and staff carry on as they did before LLDTEK. Nothing in your calendar is changed by a pause.
#After the first week
Once the current scope stays clean, routine runs stop needing attention and the daily review becomes a weekly look. New channels, services, locations or agents are added one at a time, each tested as a new surface.
You keep a one-page record of every live job: channels, hours, permissions, stop conditions, handoff owners and the pause procedure.