Building Agents at Home: Homeschooling, Parenting and More | The a16z Show
Parents with young children can now build sophisticated AI systems without sitting at a computer for hours. The key is agents that work while you're with your kids—use voice notes to assign tasks, let agents code and build during naptime, then review progress in minutes. Start by creating one 'respo
54mKey Takeaway
Parents with young children can now build sophisticated AI systems without sitting at a computer for hours. The key is agents that work while you're with your kids—use voice notes to assign tasks, let agents code and build during naptime, then review progress in minutes. Start by creating one 'responsive' agent that delegates complex tasks to specialized agents it spawns itself. This shifts your role from hands-on builder to strategic director, unlocking the ability to build technical projects while remaining present with your family.
Episode Overview
Jesse, a former YC founder and mother of four children under five, shares how she uses AI agents to homeschool her children while building sophisticated technical projects. She explains her journey from being resigned to not building 'hard things' for years to now creating better work than ever before—all while spending most waking hours with her children. The conversation covers her practical systems for homeschooling with AI, managing multiple autonomous agents, and the future of AI-assisted parenting.
Key Insights
The Confetti Time Revolution: Building During Fragmented Moments
Jesse discovered that AI agents enable building during 'confetti time'—the 10-15 minute fragments scattered throughout a parent's day. Rather than needing 8-hour focused work sessions, she assigns tasks via voice notes and lets agents execute while she's with her children. This represents a fundamental shift in who can build technical projects: parents, particularly mothers in early childhood phases, can now create sophisticated systems without traditional desk time.
Responsive Delegation: Keep Your Main Agent Unbusy
Jesse's core insight is keeping her primary agent (Sylvie) deliberately underloaded with few scheduled tasks. When Sylvie receives work requiring more than a few minutes, she delegates to other specialized agents rather than becoming bogged down. This architecture ensures instant responsiveness—critical when working in short bursts between parenting demands. The principle mirrors good management: your direct report should be available, not buried in execution.
Benevolent Neglect as a Parenting and Productivity Strategy
Jesse intentionally builds her children's 'tolerance' for unstructured time without constant stimulation, gradually increasing from 5 minutes to over 2 hours of independent play among her three older children. This serves dual purposes: it develops resilience and creativity in children while creating productive windows for AI work. She uses timers to track progress and physically removes herself rather than saying 'don't talk to me,' letting children naturally build independence.
Context-Rich AI Training: Feed Philosophy, Not Just Prompts
Jesse feeds her homeschool agent the full text of curriculum books (via PDFs or photos), her educational philosophy (via voice notes), and photos of all physical materials she owns. The agent then generates complete lesson plans including which Montessori beads to pull from which cabinet. This context-rich approach produces remarkably tailored outputs—the agent knows not just what to teach, but how Jesse wants to teach and with what specific tools.
Voice Notes Over Video: Optimizing for Token Efficiency
While video recording entire lessons seems intuitive, Jesse found that photos plus 30-second voice notes serve the same documentation purpose at far lower token cost. LLMs excel at language processing; making them 'watch' video burns tokens unnecessarily. For screen-based work (like Synthesis math), she uses Loom for transcription quality, but defaults to voice for most logging. The lesson: match your input format to what LLMs process most efficiently.
Notable Quotes
"I was resigned to not challenging myself to build technical or hard things for like the next 5 years or so. I really want to be present with my kids. I need to take this break. Basically, that is no longer true."
"A weird superpower of mine is just how incredibly motivated I am for agents to do work for me. I got my agents to learn how to build other agents on their own. So, I could be like, 'We need another agent, you guys.' and they actually can spin them up without me touching the machine, which is a little crazy."
"I try to ignore the children. I try to make sure that they're going to survive the ignoring. I step away from them and try to just see what they do. We're up to like they can they will spend more than two hours interacting and doing stuff before they come back to me."
"The first few weeks were very rough. It would be a level of pain that I wouldn't want an average person to go through. But the thing is, Jesse, it is so fantastic."
"When they spun up their own agent for the first time, none of that time is wasted. They give the agent all of our team docs, all of the contacts on myself and my husband, our children's lives. The new agent knows all of that. I don't have to feed any information. And I didn't have to ask them to do that. They knew that that would be valuable."
Action Items
-
1
Start With Voice-First Workflows
Begin documenting your daily tasks via quick voice notes (under 30 seconds each) describing what you did, what worked, and what didn't. Use these as inputs to AI tools like Claude or ChatGPT to generate logs, summaries, or next steps. This builds the muscle of delegating through natural language rather than typing, which is essential for fragmented 'confetti time' productivity.
-
2
Build Your Child's Independent Play Tolerance
Set a timer and gradually increase the duration your children play without seeking your attention. Start with 5 minutes and work up to 30+ minutes. Physically remove yourself from their space rather than saying 'don't interrupt me.' Use this time for focused work or AI experimentation. Track progress weekly to see improvement in both their independence and your available work windows.
-
3
Create a Context Repository for Your AI Agent
Gather the 'knowledge base' for any area where you want AI assistance: take photos of physical materials you own, save PDFs of key reference documents, and record voice notes explaining your philosophy or preferences. Feed this to your AI tool in one session. This upfront investment dramatically improves output quality, as the AI will reference your specific context rather than generic advice.
-
4
Implement the Delegation-First Architecture
When using AI tools, ask your primary assistant to identify tasks that would take more than a few minutes and delegate them to specialized 'sub-agents' or separate chat threads. This keeps your main interface responsive for quick questions while deeper work happens in parallel. Review the principle: your most-used AI should never be too busy to respond instantly.