# How to Tune the Copilot

Sometimes, the Copilot may struggle with identifying particular data within the spreadsheets or the connected data sources. To address such issues, you can supply the Copilot with contextual hints and there would be two ways of doing so. Let’s dive deeper into this.

#### Fine-tuning the prompt to Copilot

The first way would be to include the additional details in your prompt for the AI. Below is the example spreadsheet, which contains the agency’s partner’s data and comprises the following columns

| Partner name       | Partner Type | Partner Account Manager | Sales Manager | Status      |
| ------------------ | ------------ | ----------------------- | ------------- | ----------- |
| Innovate Solutions | Reseller     | John Doe                | Jane Smith    | Active      |
| TechConnectors     | Agency       | Sarah Johnson           | Mike Williams | Pending     |
| DataInnovations    | Integration  | Alex Turner             | Emily Davis   | In Progress |
| Digital Dynamics   | Reseller     | Mark Anderson           | Jessica Brown | Active      |
| CloudTech Alliance | Agency       | Laura Miller            | Brian Taylor  | Lost        |

**Example 1**

Let’s imagine we want to use the abbreviation PAM (\*partner account manager) in our prompt for the AI and our prompt looks something like this:

*Who is the PAM for Open AI?*

In this scenario, the Copilot may lack the understanding what’s the meaning of PAM and, hence, will fail to provide the correct reply to this question

Let’s try modifying this query so that the AI would have an additional context:

a. Who is the Partner Account Manager for Open AI? b. Who is the PAM for Open AI? (PAM = Partner Account Manager) c. Who is the Partner Account Manager for partner Open AI?

#### Example 2

In this example, we want to pull the partner account manager for the company called ‘X X Corp Inc.’ *Who is the partner account manager for X?*

In this scenario, Copilot may struggle with matching ‘X’ with the companies listed in our source data sheet. Let’s modify our prompt, by adding some contextual hint

*Who is the partner account manager for X? (Partial names accepted)*

#### Example 3

In this example, we want to pull what projects one of the partner managers (Nick) handles

*What does Nick handle?*

In the scenario, the Copilot may respond with the request for additional clarifications, regarding Nick. Let’s modify the original prompt, adding the contextual hint:

*What partners does partner account manager Nick handle?*

#### Example 4

It's important to remember that it relies on data from your apps and tables. The AI's responses are only as good as the information you provide. To make the most of Copilot, ask specific, data-focused questions

Let’s say we want the AI to pull the data, related to Facebook Ads, using this query

*How good are my Facebook Ads?*

As you can see, this query lacks the details and the AI fails to respond with a quality reply 

<figure><img src="/files/AWSwXfsoJWApFtta0pIg" alt=""><figcaption></figcaption></figure>

Now, let's try this one *Retrieve the click-through rate (CTR) data from my Facebook Ads campaign, targeting the US. Please break it down by months*&#x20;

<figure><img src="/files/6GQn6gAuWN0bO7ZoUhrL" alt=""><figcaption></figcaption></figure>

#### Configuring custom instructions

You can make your contextual hints persistent, by adding them to the 'Instructions' section on the Copilot chat tab. Just click on this button and enter your custom contextual hints&#x20;

<figure><img src="/files/p6ZorvvJxMsQFdVMWd8D" alt=""><figcaption></figcaption></figure>


---

# 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://docs.getconduit.app/copilot/how-to-tune-the-copilot.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.