Gemini 4 Argon Really Surprised Me WebTwo Minute PapersOct 19 Google Gemini 4 agents just beat Minecraft with no human intervention WebEngadget AIOct 19 Who could it be? 🧐 OfficialGoogle on YouTubeOct 18 Customize your Gemini chats experience Learn & DoGemini Help CenterOct 17 Gemini 4 Argon Made a Fall Guys Clone WebMatthew BermanOct 17
Google OfficialWebLearn & DoResources People Podcasts
All NewsTradeVideosVoicesCommunityResearchPodcasts
Subvolts take Build a private AI assistant by splitting tasks between models and encrypting data locally. Sep 23, 2025
For you
Developers
Learn how to architect and implement privacy-preserving AI assistants by splitting workflows and using local encryption.
Business
Understand the trade-offs and strategies for integrating AI into consumer products while maintaining user data security.
Hype note
3/5Notable

This episode offers a practical case study on building a privacy-focused AI assistant, relevant to many developers and businesses.

Practical AI
Practical AI · Ep 329 Episode

Creating a private AI assistant in Thunderbird

53 min Interview Hosts: Daniel Whitenack, Chris Benson
0:0053:08

Audio streams from Daniel Whitenack and Chris Benson. Your place is remembered on this device.

Go to the podcast episode page
Open Practical AI in
Apple PodcastsOvercastPocket Casts
Follow the show
Apple PodcastsSpotifyOvercastPocket CastsAny Android app

From Subvolts

Summary and analysis by Subvolts editorial, AI-assisted.

Summary

In this Practical AI episode, hosts Daniel Whitenack and Chris Benson interview Chris Aquino, a software engineer at Thunderbird. They discuss the development of a privacy-preserving AI executive assistant for Thunderbird. The conversation covers key design decisions like remote inference, local encryption, and selecting appropriate AI models to split workflows.

Explained simply

This episode explains how Thunderbird built a smart assistant that respects your privacy. Instead of sending everything to big AI companies, they use smaller, specialized AI tools and keep your data safe on your computer.

Why listen

This episode details the engineering challenges and solutions for building a private AI assistant. Listeners will learn about balancing privacy with AI functionality, model selection strategies, and workflow decomposition. It is for developers and product managers interested in integrating AI into existing applications while prioritizing user data security. This work contrasts with frontier labs' focus on large, centralized models, highlighting a growing need for more private, distributed AI solutions.

On the mic
  • HostDaniel Whitenack
  • HostChris Benson
  • GuestChris Aquinosoftware engineer, Thunderbird
About
GoogleAgentsDeveloper ToolsSafety
Key facts
  • Episode Title: Creating a private AI assistant in Thunderbird
  • Guest: Chris Aquino, software engineer at Thunderbird
  • Show: Practical AI
  • Length: 53 minutes
  • Key Design Decisions: Remote (but confidential) inference, local encryption, model selection
Key moments and transcript

Chapters from the publisher ↗ · Transcript ↗ Key moments and transcript search come with a later piece.

Show notes from Daniel Whitenack and Chris Benson

In this episode, Daniel and Chris are joined by Chris Aquino, software engineer at Thunderbird to hear the story of how they developed a privacy-preserving AI executive assistant. They discuss various design decisions including remote (but confidential) inference, local encryption, and model selection. Chris A. does an amazing job describing the journey from "let the big LLM do everything" to splitting apart the workflow to be handled by multiple models. 

Featuring:

Links: 

Sponsors:

  • Shopify – The commerce platform trusted by millions. From idea to checkout, Shopify gives you everything you need to launch and scale your business—no matter your level of experience. Build beautiful storefronts, market with built-in AI tools, and tap into the platform powering 10% of all U.S. eCommerce.
    Start your one-dollar trial at shopify.com/practicalai

Upcoming Events: 

  • Join us at the Midwest AI Summit on November 13 in Indianapolis to hear world-class speakers share how they’ve scaled AI solutions. Don’t miss the AI Engineering Lounge, where you can sit down with experts for hands-on guidance. Reserve your spot today!
  • Register for upcoming webinars here!
Jargon decoded
inference
Chris Aquino
using a trained AI model to make predictions or generate output
LLM
Chris Aquino
Large Language Model, a type of AI trained on vast amounts of text data
Published
Show
Practical AI · Daniel Whitenack and Chris Benson
Length
53 min
Companies
Google
Products
Thunderbird
Summary by
Subvolts, using an AI model (how we work). Spotted a mistake? Tell us.
#Agents#Developer Tools#Safety

Best of Practical AI

More on Google this week

Questions people ask

What is Thunderbird developing?
Thunderbird is developing a privacy-preserving AI executive assistant. This assistant aims to handle tasks while keeping user data secure.
What design decisions were made for the AI assistant?
Key decisions included using remote but confidential inference, implementing local encryption, and carefully selecting AI models to split workflows.

Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Daniel Whitenack and Chris Benson.