Building & improving agents

View as Markdown

Building a production agent is a loop, not a one-shot. This page lays out the path from a first draft to a deployed agent you keep improving with real traffic.

1

Build the first version

Start with a single-state agent: a clear instruction, a voice and language, and only the tools the core flow needs. Resist adding states or tools you can’t yet test.

2

Review & test the prompt

Test every change before it ships:

  • test agent: talk to the agent live and listen for turn-taking, language, and tone.
  • Simulate: run scripted / simulated conversations to check behaviour repeatably.
3

Set up outcome variables

Define the output variables that capture what happened, disposition, resolution, collected fields, and the goal that judges success. These are what you’ll measure and improve against.

4

Deploy

Ship the agent to a channel, telephony, a web widget, or code, deliberately, from a tested config version.

5

Set up key metrics & dashboards

Wire up Agent Analytics and build the boards you’ll watch: connectivity, engagement, outcomes, and the business metric that actually matters.

6

Improve

Use call logs, transcripts, and outcome metrics to find where the agent fails, tighten the prompt or tools, re-test, and roll forward. Repeat.

Goals

A goal is the success criterion for an agent. It powers automated conversation evaluation against both live and test agent traffic, and its results feed the Goals dashboard in Agent Analytics.

You can set one goal per agent. Keep it focused on the single business outcome that defines success; use Agent Analytics for broader operational metrics.

Rule-based conditions

A goal is built from rules that check conversation outcomes - for example whether a required variable was set, whether a disposition matches an expected value, or whether a numeric score clears a threshold.

OperatorMeaning
equalsExact match
not_equalsNot equal
inValue is in a set
gtGreater than
ltLess than

Additional operators may appear in the dashboard as the evaluation system expands.

See the results

Once a goal is configured, every connected call is evaluated against it. The outcomes appear in the Goals tab under Agent Analytics - achievement rate, achieved-vs-not cohorts, and the patterns that correlate with success.