# Tips & tricks for your agent

A sipgate AI Agent is quick to set up and handles calls reliably. To ensure your AI phone assistant responds optimally in all situations and delivers a professional conversation experience, these best practices help with setup and testing. Real phone conversations are diverse: callers phrase things differently, switch topics spontaneously, or get straight to the point. Your AI Agent should handle all these scenarios with confidence.

In this article, you will find practical tips for optimizing your AI Agent: from the perfect greeting and realistic test scenarios to targeted checks of call forwarding. This ensures that your AI phone assistant not only works technically, but also convinces callers with natural, helpful responses.

### **Tip 1: Keep the greeting short**

The first few seconds matter. A good greeting is clear and concise. The agent states its name and explains how it can help. More than that is not needed for a good start.

{% hint style="info" %}
Make sure to phrase the greeting in spoken language. The text will be read aloud exactly as it is written here.
{% endhint %}

In certain cases, the greeting can also be used to point callers to a topic or to ask a simple introductory question right away.

Pay attention to this:

* Instead of "Good day, this is the sipgate Agent. I'm here to help you with your concerns," better: "Hello, this is Lisa, the digital assistant from sipgate Support. How can I help you?"
* Do not string together multiple greeting phrases
* Get straight to the point<br>

### **Tip 2: One question at a time**

Multiple questions at once are overwhelming. Ask them one after the other.

Pay attention to this:

* Not "What is your name and what is it about?" but first "What is your name?" and then "What is it about?"
* Each question should address only one piece of information

Short tests by colleagues or external people often provide more valuable feedback than your own runs.

### Tips for testing

#### Tip 1: Deliberately vary how calls start

Callers rarely start with a clear, perfectly phrased request. So test different openings, such as very short greetings from callers, direct concerns, or uncertain starts.

Pay attention to this:

* Does the agent reliably recognize the request?
  * If necessary, supplement the agent's greeting to give callers guidance on how to phrase their request, or formulate the first task in the [Playbook](/ai-agents/en/behavior/playbooks.md) so that the agent knows what type of request to expect. The agent is able to help callers if it cannot assign the stated request to a category.
* Does the start remain calm and understandable?
* Does the agent not steer the conversation too early?

#### Tip 2: Test different phrasings

The same request can be phrased in many ways. Test synonyms, paraphrases, colloquial language, or even slightly incorrect terms.

Pay attention to this:

* Does the agent always lead to the same result?
* Does it ask follow-up questions sensibly, or does it react unnecessarily?

#### Tip 3: Include real conversation situations

Background noise, interruptions, or very fast speech are part of everyday life. Deliberately include such situations in the tests.

Pay attention to this:

* Does the agent remain understandable?
* Does it respond in a way that makes sense?
* Does it lose the thread?

#### Tip 4: Review conversations afterward

Look at test conversations later from the caller's perspective. This quickly shows whether answers are too long, unclear, or redundant.

{% hint style="info" %}
Test conversations can be reviewed in the transcripts of the [analysis in the logged-in area](https://app.sipgate.com/frontdesk-analysis?) .
{% endhint %}

Pay attention to this:

* Is the conversation flow logical?
* Are answers clear and not too verbose?
  * A common cause of verbose answers is overly long responses in [customer questions](/ai-agents/en/behavior/kundenfragen.md) in this case, shorten the corresponding response.

#### Tip 5: Test call forwarding specifically

When calls are handed over, the transition should work smoothly. Deliberately test call forwarding multiple times.

Pay attention to this:

* Is it clearly communicated what happens next?
* Does the handover also work outside business hours?

#### Tip 6: Test the agent's limits

An agent does not need to be able to solve every request. What matters is how it handles situations that it cannot cover.

Pay attention to this:

* Does the agent communicate its limits clearly?
* Does it offer sensible alternatives or transfers?

#### Tip 7: Get external tests

Short tests by colleagues or external people often provide more valuable feedback than your own runs.

Helpful questions afterward:

* What was unclear?
* Where did you get stuck?
* What worked well?

#### Tip 8: Test again after changes

After adjustments to knowledge, scenarios, or conversation flow, short tests should be part of the process. 3–5 test calls after each major change have proven effective.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://help.sipgate.de/ai-agents/en/behavior/tipps-and-tricks-fur-ihren-agent.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.