OpenAI Evals Explained with Examples | AI Voice

In this video, we break down how you can evaluate the performance of AI models like GPT using real-time data, without the need for third-party tools. Learn how to analyze chat completions, track KPIs, reduce hallucinations, and streamline your AI workflow-all directly within OpenAI’s platform. Whether you're in voice AI, customer service, or any field using AI-generated data, this tool is a game changer.
My resource hub:
hub.integratic...
Work with us 👋🏼
integraticus.com/
My Links 🔗
👉🏻 Subscribe: / @jannismoore
👉🏻 Instagram: / jannismoore
👉🏻 LinkedIn: / jannismoore
👉🏻 More ways to reach me: integraticus.c...

[@biiiiiimm]

Very interesting ! Im working on adding this kind of feature in my SaaS, i was a bit confused about how it works exactly, you helped me a lot ! Good job

[@premium2024-gt3bn]

excellent channel. thanks

[@bulutosman]

is not there anyway to evaluate function callings?

[@jannismoore]

You can always custom-build something and Vapi also offers its own structured date format that you can use to get KPIs.
If you really want to analyze only the tool calling responses, you can probably do that by simple looping through the messages array within the end of call reports while fetching the tool call responses

[@Zack-AP]

Very sick video man. Did you create that replit for the vapi data with the replit agent by chance?

[@jannismoore]

I used it for the foundation - then it’s just optimizing the code for my needs.
It’s not always getting things right, so there’s quite a bit more of manual work needed, but it definitely shortens the dev time. 👍🏻

[@manideepagunde1632]

How this gonna help in the web scrapping sir

[@jannismoore]

You can analyze your scraped data if you look for something specific. I guess you could use it also for extraction with the response feature they offer