Early childhood education app (AI-first)¶
| Field | Value |
|---|---|
| Status | Active — team product (2026-06-14) |
| Type | Product / edtech — AI-first early childhood education |
| Team | Small team (Darcy + collaborators) |
| Priority class | Top priority — product building (see What-Im-Working-On, index) |
Description¶
Building an AI-first early childhood education app with a small team. The product target is learning experiences for young children where AI is native to the design — not a bolt-on chat box.
Parallel meta-goal: learn how to reinvent the nature of building and collaboration when the team includes both humans and non-human agents (LLMs, coding agents, review bots, scheduled jobs, etc.). This project is where Darcy practices human + agent co-building at team scale — complementary to the solo Crusty / Hermes fleet on projects/crusty and the personal wiki pattern on projects/llm-maintained-context.
What “AI-first” means here¶
- Product shape driven by what models can reliably do for young learners and caregivers — not retrofitted onto a static-content app.
- Build process treats agents as first-class collaborators: spec → draft → human review → ship, with explicit gates where child-facing or trust-sensitive surfaces need human judgment.
- Team workflow experiments with who (or what) owns which slice — design, code, content, QA, parent-facing copy — and how context stays shared across human and agent contributors.
Wiki posture¶
- Public wiki: product direction, collaboration patterns, and cross-links to the personal agent fleet — no employer/client names, repo URLs, API keys, or unreleased feature detail unless Darcy explicitly ingests them.
- Prior art: an earlier wiki page (
education-startup-ux, removed 2026-06-06) covered overlapping day-job context; this page is the curated top-priority product line as of 2026-06-14.
Thesis fit (goals)¶
| Leg | How this project strengthens it |
|---|---|
| Context depth | Shared team context (specs, child-development constraints, iteration history) must survive across humans and agents |
| Trusted access | Child-facing product = high trust bar; agents assist build and ops, not unchecked child interaction without gates |
| Durable trust | Parents/caregivers and teammates must trust the process — same draft → review → ship discipline as projects/crusty |
Related¶
- projects/crusty — solo agent fleet; Telegram human-gate pattern
- projects/openclaw-autonomy-org — multi-agent umbrella framing
- projects/llm-maintained-context — wiki-as-shared-memory meta-skill
- projects/ai-model-distillation — on-device / small-model path (historical Boxy POC informed distillation Phase 1)
- projects/triviabalance — another learning/engagement surface (seniors, web)
- projects/happy-body — coaching trust surface (life & identity, not this bucket)
Current posture (2026-06-14)¶
- Active team build — top-priority product building alongside Crusty and Operator Agent creative lane.
- Learning in public (wiki-safe): document collaboration patterns (human + agent roles, review gates, what agents are allowed to touch) as they stabilize — not proprietary product specs.