<!-- START doctoc generated TOC please keep comment here to allow auto update -->
<!-- DON'T EDIT THIS SECTION, INSTEAD RE-RUN doctoc TO UPDATE -->
**Table of Contents** *generated with [DocToc](https://github.com/thlorenz/doctoc)*
- [Example: Using LLMs to Extract Key Financial Data for Earnings Call Transcript Analysis](#example-using-llms-to-extract-key-financial-data-for-earnings-call-transcript-analysis)
- [Prerequisites](#prerequisites)
- [Example Code](#example-code)
<!-- END doctoc generated TOC please keep comment here to allow auto update -->
# Example: Using LLMs to Extract Key Financial Data for Earnings Call Transcript Analysis
> [!NOTE]
> Imagine reading a lengthy earnings call transcript to extract key data mentioned by the CEO, CFO, or analysts, which could take at least 10 minutes. Using an LLM, this can be done in just a few seconds.
>
> This document provides an example of how to use the large language model to summarize key financial data from earnings call transcripts.
## Prerequisites
To run this example, you need:
1. An OpenAI-compatible API key (`OPEN_AI_API_KEY`)
2. A LLM model with a function call tools capability.
3. A LLM model with a thinking capability.
Our tests show that even free, small-parameter models(e.g. [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B)) can deliver excellent results.
> [!TIP]
> You can obtain a free API for small-parameter models from [SiliconFlow's Chinese website](https://www.siliconflow.cn/pricing).
>
> For higher accuracy and performance, consider using larger models from [OpenAI](https://openai.com/index/openai-api/), [DeepSeek](https://api-docs.deepseek.com/), [QWen](https://qwen.ai/apiplatform), or [Gemini](https://ai.google.dev/gemini-api/docs).
## Example Code
Below is an example demonstrating how to fetch key financial metrics from earnings call transcripts using LLM model.
```python
from openai import OpenAI
from defeatbeta_api.data.ticker import Ticker
from defeatbeta_api.client.openai_conf import OpenAIConfiguration
# Initialize the Ticker
ticker = Ticker("AMD")
# Fetch earnings call transcripts
transcripts = ticker.earning_call_transcripts()
# Configure the OpenAI client
llm = OpenAI(
api_key="OPEN_AI_API_KEY", # Replace with your OPEN_AI_API_KEY
base_url="OPEN_AI_API_END_POINT" # Replace with your OPEN_AI_API_END_POINT
)
# Summarize key financial data for Q2 2025 with llm
res = transcripts.summarize_key_financial_data_with_ai(
2025,
2,
llm,
OpenAIConfiguration(model='Qwen/Qwen3-8B'))
print(res.to_string())
```
---
```text
symbol fiscal_year fiscal_quarter speaker paragraph_number key_financial_metric time_scope value currency_code
0 AMD 2025 2 Lisa T. Su 3 total_revenue this_quarter 7.700000e+09 USD
1 AMD 2025 2 Jean X. Hu 4 gaap_gross_margin this_quarter 4.300000e-01 USD
2 AMD 2025 2 Lisa T. Su 3 non_gaap_gross_margin this_quarter 5.400000e-01 USD
3 AMD 2025 2 Jean X. Hu 4 gaap_operating_expense this_quarter 2.400000e+09 USD
4 AMD 2025 2 None None non_gaap_operating_expense this_quarter NaN None
5 AMD 2025 2 Jean X. Hu 4 gaap_operating_income this_quarter 8.970000e+08 USD
6 AMD 2025 2 None None non_gaap_operating_income this_quarter NaN None
7 AMD 2025 2 Jean X. Hu 4 gaap_operating_income_margin this_quarter 1.200000e-01 USD
8 AMD 2025 2 None None non_gaap_operating_income_margin this_quarter NaN None
9 AMD 2025 2 None None gaap_net_income this_quarter NaN None
10 AMD 2025 2 None None non_gaap_net_income this_quarter NaN None
11 AMD 2025 2 None None ebitda this_quarter NaN None
12 AMD 2025 2 None None adjusted_ebitda this_quarter NaN None
13 AMD 2025 2 Jean X. Hu 4 gaap_diluted_earnings_per_share this_quarter 4.800000e-01 USD
14 AMD 2025 2 None None non_gaap_diluted_earnings_per_share this_quarter NaN None
15 AMD 2025 2 Lisa T. Su 3 fcf this_quarter 1.200000e+09 USD
16 AMD 2025 2 Jean X. Hu 4 total_cash_position this_quarter 5.900000e+09 USD
17 AMD 2025 2 Jean X. Hu 4 share_repurchase this_quarter 4.780000e+08 USD
18 AMD 2025 2 None None capex this_quarter NaN None
19 AMD 2025 2 Jean X. Hu 4 total_revenue_forecast next_quarter 8.700000e+09 USD
20 AMD 2025 2 None None gaap_gross_margin_forecast next_quarter NaN None
21 AMD 2025 2 Jean X. Hu 4 non_gaap_gross_margin_forecast next_quarter 5.400000e-01 USD
22 AMD 2025 2 None None gaap_operating_income_margin_forecast next_quarter NaN None
23 AMD 2025 2 None None non_gaap_operating_income_margin_forecast next_quarter NaN None
24 AMD 2025 2 None None gaap_operating_expense_forecast next_quarter NaN None
25 AMD 2025 2 Jean X. Hu 4 non_gaap_operating_expense_forecast next_quarter 2.550000e+09 USD
26 AMD 2025 2 None None ebitda_forecast next_quarter NaN None
27 AMD 2025 2 None None adjusted_ebitda_forecast next_quarter NaN None
28 AMD 2025 2 None None gaap_earnings_per_share_forecast next_quarter NaN None
29 AMD 2025 2 None None non_gaap_earnings_per_share_forecast next_quarter NaN None
30 AMD 2025 2 None None capex_forecast next_quarter NaN None
31 AMD 2025 2 None None total_revenue_forecast full_fiscal_year NaN None
32 AMD 2025 2 None None gaap_earnings_per_share_forecast full_fiscal_year NaN None
33 AMD 2025 2 None None non_gaap_earnings_per_share_forecast full_fiscal_year NaN None
```
---
> [!IMPORTANT]
> For key financial metrics labeled as `xxx_forecast`, the original text often provides a forecast range for the next quarter. In such cases, the model will output the midpoint of that range. Note that this value may not be explicitly stated in the transcripts.
>
> In contrast, metrics not labeled as `xxx_forecast` will be extracted directly from the transcripts without any calculation.
>
> If a metric is not mentioned in the transcripts, return null for that field.
