Every Vibe Code Camp
Talk 03

Personal AI suite

Ashe Magalhaes

Summary

Overview

Ashe Magalhaes, former founder of Hearth AI and ML engineer with experience at Apple and in politics, presents her philosophy and approach to building a personal AI suite. The session demonstrates her "Ashe AI" system—a comprehensive personal infrastructure where multiple specialized agents help manage her daily life, relationships, and creative output.

Key Themes

1. Relational Intelligence as Core Philosophy

Ashe's fundamental thesis centers on "relational intelligence"—the idea that AI should augment human experience by extending our ability to reason about who we're connected to, why, and how to act across the "optimization landscape" of all our relationships. This philosophy emerged from her work in politics, philanthropy, and tech, where she consistently solved the same problem: mining and matching people within massive networks.

2. Personal AI Suite Architecture

The Ashe AI suite represents a unified personal infrastructure with:

  • Multiple specialized agents with different context windows and focuses
  • Various input channels: Slack, email, text messaging
  • Integrated workflows connecting goals, content creation, and relationship management
  • Public and private layers: Some pages are public-facing (ash.ai), while others remain private

3. Video as Authentic Communication

In an era of AI-generated content "slop," Ashe argues for video as a medium that preserves authentic human connection. The stutters, imperfections, and human presence in video create genuine connection that text and AI-generated images cannot replicate. For 2026, she's focused on getting comfortable on camera and systematizing video production.

4. Network Visualization Challenges

A recurring theme is the difficulty of visualizing relationship networks effectively. Ashe notes that neither humans nor LLMs excel at network visualization in 2D, pointing to spatial computing and 3D representations as the eventual solution. Her "neural dendrite" co-occurrence network visualization represents her latest attempt at this challenge.

Main Demonstrations

  • Ashe AI Dashboard: Hidden URLs and secret pages containing projects and agents
  • Stream/Quote Capture: A public mood board where ambient thoughts are captured via Slack messages to agents
  • Rolodex System: Personal relationship management pulling from intentional notes rather than integration noise
  • Contribution Graphs: GitHub-style tracking for personal goals tied to agent accountability
  • Remotion Video Experiments: Live demonstration of prompting video generation (with mixed results)

Session Flow

  • Introduction and background (Hearth AI, relational intelligence thesis)
  • Ashe AI suite walkthrough
  • Network visualization discussion
  • Philosophy on video authenticity in AI era
  • Live Remotion experimentation
  • Discussion of reading/multitasking while waiting for agents
  • Wrap-up and contact information

Practical Takeaways

  • Build incrementally: The first 20 versions will likely be "shitty"—embrace the creator mindset
  • Use agents for accountability: Connect personal goals to agent tracking and contribution graphs
  • Intentional relationship data: Manual notes on people create more meaningful context than automated integrations
  • Multiple agent architectures: Different agents with different focuses (OpenAI tool calling for simple tasks, custom architectures for complex ones)
  • Pair models for debugging: Using Opus with Codex can be effective—slow but precise

Speaker Background

  • CS major, former FANG engineer
  • ML engineer in politics
  • Worked at Apple abroad
  • Worked in philanthropy
  • Founder of Hearth AI
  • Currently exploring next steps
  • Twitter: @AsheBytes

Key Concepts

Relational Intelligence

AI that augments human experience by extending our ability to reason about who we're connected to, why those connections matter, and how to navigate the "optimization landscape" of all our relationships.

Personal AI Suite

A unified infrastructure of multiple specialized AI agents with different context windows, input channels, and focuses—all connected to a single backend for cross-pollination.

Co-occurrence Networks

A network visualization approach where one entity (you) sits at the center and all other nodes are positioned based on their relationship to you, rather than their relationships to each other.

Contribution Graphs for Personal Goals

GitHub-style activity tracking applied to personal habits and goals, showing patterns of consistency over time rather than streak counts.

Intentional vs. Integration-Based Relationship Data

The distinction between relationship data that comes from automated integrations (emails exchanged, calendar events) versus data that comes from dedicated thought and conversation.

Notable Quotes

"My fundamental thesis is around relational intelligence. It's this idea that AI should augment the human experience by extending our ability to reason on who we're connected to and why and how you act and take a step across your relationship, across the optimization landscape that is all of your relationships."
"These were all the same problems. It was, okay, there are these massive networks to mine and match people with large amounts of funding or opportunity or other people. And how do you do that effectively over time?"
"I've really come to believe through the Hearth journey is that integrations can add unnecessary noise to a Rolodex and actually the people that are top of mind or that I care about are people that I'm naturally thinking about, meeting with live, adding notes on."
"I've dedicated thought or like conversational back and forth to these people. They're beyond like the integrations of we've exchanged an email. It's like, oh, I've dedicated thought."

Tools Mentioned

Claude CodeCursorCodex 5.2 (OpenAI)OpenAI Tool CallingRemotionRemotion StudioSlackEmailText MessagingTwitter/XNext.jsGoogle GDELT ProjectTikTokAshe AIRolodex Agent Architecture

Transcript

03 of 18
← Previous Non-technical builder Next → Infinite loop agent