playnext.fm

.NET Rocks! · Carl Franklin

Local AI Models with Joe Finney

October 2, 2025·55 min·5 clips
Joe Finney reveals why local LLMs are slower than cloud models and the hidden costs of context windows.
It starts with the usual show ritual. A quick opening leads into Carl, Richard, and Michelle LaRue Bustamante talking through Cybersecurity Intersection next to Dev Intersection. Then the episode settles into local AI. Joe and the hosts push back on treating AI and LLMs as the same thing, and the hosts clearly like that distinction. The examples stay concrete: OCR, image segmentation, image detection, and object detection. Hugging Face becomes the doorway, with the hosts pausing on it as a huge online place to find and download models. Then comes the catch. Downloading a model is not like running a program. A curious developer still needs something that can talk to it. Microsoft gets a practical mention here, with AI Dev Gallery framed as a playground for trying models without pretending the model is the whole app. The tone stays loose and technical. One host keeps pointing back to the wider AI tooling that existed before ChatGPT took over the conversation. Joe keeps it grounded. Local AI comes across as a spread of useful workloads, not just cloud chatbot culture shrunk down. There is some uncertainty around when a model needs fine-tuning or is not already ready to use. By the end, the hosts sound satisfied. They got the local LLM and local AI conversation they wanted with Joe.

As heard by us

A practical reset: local AI is broader than chatbots and easier to approach than the hype suggests.

Joe Finney gives Richard and Carl a practical reset on local AI. The conversation is clearest when it moves past chatbot talk and into the wider model stack, with Hugging Face framed as a large repository and real tasks like OCR, image segmentation, image detection, and object…

Read the full review in PlayNext →

Why you'd press play

You want a grounded look at local AI that goes past chatbot talk.

Read the full recommendation in PlayNext →
Listen to the show on
Apple PodcastsSpotify
Play clips from this episodeMore from .NET Rocks!