Schedule at a glance
- Tue / Thu — 11:00 AM-12:00 PM ET / 8:00-9:00 AM PT (1 hour)
- Sat / Sun — 2:00-4:00 PM ET / 11:00 AM-1:00 PM PT. First hour is the required build sprint; second hour is optional office hours, except the final Sunday which is Demo Day.
- Cohort 1 dates: Tue Jun 16, Thu Jun 18, Sat Jun 20, Sun Jun 21, Tue Jun 23, Thu Jun 25, Sat Jun 27, Sun Jun 28, 2026
- Cohort 2 dates: Tue Jul 7, Thu Jul 9, Sat Jul 11, Sun Jul 12, Tue Jul 14, Thu Jul 16, Sat Jul 18, Sun Jul 19, 2026
Daily structure
Weekday session (1 hour)
- 0:00-0:10 — Demo. Instructor walks through today's concept on a real example.
- 0:10-0:45 — Guided build. Students build alongside, instructor floats.
- 0:45-1:00 — Share + live Q&A. Screen-shares, unblockers, what's next.
Weekend session (2 hours)
- 2:00-3:00 PM ET / 11:00 AM-12:00 PM PT — Required build sprint. Guided start, then independent build with live instructor help.
- 3:00-4:00 PM ET / 12:00-1:00 PM PT — Optional office hours for debugging, code review, and bonus moves. Students who are green can leave after the first hour.
- Final Sunday exception: Demo Day uses the full 2-hour block for demos, awards, and closing.
Tool stack
Required for everyone (parent-owned accounts)
- Replit Core plan (~$20/mo, parent-paid, can cancel after camp) — IDE, agent, deployments, and scheduled jobs in one browser tab
- OpenAI API account (parent-owned, parent-managed key with a $10/mo hard usage cap) — powers the AI inside everything students build. Camp does not provide API keys or cover usage costs.
- GitHub account (free, 13+) — for pulling down scaffolding repos we provide
- Telegram or Discord account (13+) — optional bot surface for students who take the Week 2 bot path; scheduled Replit runs are the fallback
- ChatGPT account (13+, parental consent) — for chat-style "talk to AI" lessons
Case studies (we'll learn about them, not build with them)
- Agent harnesses like Claude Cowork and OpenClaw — we use these as case studies so students understand how production agents are structured. Students build a simpler, starter-based version of the same pattern in Replit using the OpenAI API.
Day 0 — Pre-flight setup
Before Day 1 — ~15 minutes with a parent
Theme: "Show up to Day 1 with everything already working."
Every camper completes a short setup checklist the week before camp starts. A parent creates a Replit Core account (parent-owned, parent-paid), and the camper signs in, forks the starter template, and clicks Run to confirm everything works. AI access is built into Replit — no separate API keys or accounts needed. We'd rather fix a broken setup on a Sunday than have your camper sit out the kickoff.
Deliverable: Camper can log in to Replit, the account shows Core, and the starter template runs successfully. If anything fails, email us before Day 1 and we'll get on a 10-min call.
Enrolled families receive the setup guide link by email the week before camp.
Week 1 — Build something real
Day 1 (Tuesday) — Setup & first working tool
Theme: "By the end of today, something you built is running in the cloud."
Today's goal
- Sign in to Replit, confirm your OpenAI API key works inside a Repl, fork the starter scaffolding from GitHub
- Generate and run your first tool — type something in, get something AI-generated back from gpt-4o-mini, see it on screen
- Understand the basic shape of every tool you'll build: ask the user something, send it to the OpenAI API, show the result
Deliverable: A working tool running in your Replit — type something in, get something AI-generated back.
Bonus moves (if you finish early)
- Wire in a second prompt mode (e.g., "explain like I'm 10" vs "explain like I'm in college")
- Add a small bit of state so the tool remembers the last result
- Polish the UI — make it feel intentional, not generic
Tools introduced: Replit, OpenAI API (gpt-4o-mini), GitHub.
Day 2 (Thursday) — Tech spec + scaffold
Theme: "Before you build, know what you're building."
Today's goal
- Pick a project direction (AI study planner, daily personal dashboard, news / celebrity tracker, homework breakdown tool, content generator, goal planner)
- Write a tech spec (
spec.md): who it's for, the one thing it does, input → output, APIs needed - Hand the spec to Replit Agent and scaffold the project structure — architecture in place, no features yet
Deliverable: A tech spec (spec.md) plus a running scaffold — ready for design (Day 3) and the build (Day 4).
Bonus moves (if you finish early)
- Add a stretch-goals section to your spec for later features
- Map out your routes and where the AI calls will happen
Tools introduced: Prompt structure (role, context, constraint, format).
Day 3 (Saturday) — Design Inspiration (anti-AI slop)
Theme: "AI can build you anything — but first, know what you want it to look like."
Today's goal
- Define "AI slop" and learn to spot it in real products
- Gather design references from multiple sources (Coolors, Google Fonts, Variant, Pinterest, real apps) into a references folder an agent can work from
- Articulate your design direction in one sentence — no building yet
Deliverable: A design references folder with 5–10 items (palettes, fonts, app screenshots) in day-03/design/references/, plus your design vibe posted in Discord.
Bonus moves (if you finish early)
- Go deeper on references — component styles, spacing, micro-interactions from apps you love
- Write a "don't do this" list of slop your agent should avoid on Day 4
Optional office hours: 3:00-4:00 PM ET / 12:00-1:00 PM PT for peer review of references and design-direction help.
Tools introduced: Coolors, Google Fonts, Variant, Pinterest.
Day 4 (Sunday) — Design Spec + Build v1
Theme: "You've been planning. Now watch it come to life."
Today's goal
- Turn your references into a formal design spec (
design-spec.md) using Replit Agent - Hand both specs (tech + design) to Agent and have it build v1 — works like the spec, looks like the design
- Evaluate and iterate: spot what doesn't match your specs, describe the fix, let Agent fix it, evaluate again. You're directing, not coding.
Deliverable: A running v1 with intentional design built from your specs. Both specs + working app pushed to GitHub.
Bonus moves (if you finish early)
- Add memory (localStorage) so the tool remembers the user across sessions
- Tighten the design pass — spacing, states, and microcopy in your app's voice
Optional office hours: 3:00-4:00 PM ET / 12:00-1:00 PM PT for debugging and helping students whose v1 didn't land.
Tools introduced: the design-spec and build-v1 Agent Skills.
Week 2 — Make it powerful
Day 5 (Tuesday) — Feature Review + Midpoint Demos
Theme: "Before you build more, decide what matters most."
Today's goal
- Evaluate your feature ideas: "need to have" vs. "nice to have"
- Prioritize what to build next based on user value, not personal excitement
- Give a 60-second midpoint demo — show the app running with a real input and output — and give/receive product feedback
Deliverable: A completed 60-second midpoint demo and a clear direction for the feature(s) you'll build next.
Bonus moves (if you finish early)
- Write a one-line problem statement for the single feature that matters most
- Leave specific, useful feedback on two other students' demos
Tools introduced: need-vs-nice prioritization, Discord feedback threads, the 60-second demo format.
Day 6 (Thursday) — Agents + Automation
Theme: "Make your app work while you sleep."
Today's goal
- Learn the three kinds of automation: scheduled jobs, bots, and recurring tasks — you describe what you want, the AI builds it
- Use Replit Connectors (one-click OAuth) and Scheduled Deployments so your code runs on a timer even when the browser is closed
- Add one working automation to your project, or build the guided Discord homework-reminder bot
Deliverable: One working automation — a scheduled job, a bot, or a recurring task — verified by seeing it fire at least once.
Bonus moves (if you finish early)
- Flip
USE_AI = Trueso the reminder writes a fresh message each run - Wire up the Discord Connector for an interactive bot instead of a one-way webhook
Tools introduced: Replit Connectors, Scheduled Deployments, the add-automation Agent Skill, the reminder_bot.py starter.
Day 7 (Saturday) — Ethics, GTM + Ship
Theme: "Ship responsibly. Ship to real people. Ship ready."
Today's goal
- Understand AI ethics for your own app: privacy, bias, security, content safety, transparency — and run the
ethics-reviewskill on it - Learn the difference between cloud and local LLMs and why it matters
- Write a 5-second pitch and a GTM plan with the
gtm-planskill, then ship one real GTM action in class
Deliverable: ethics-review.md saved, a written GTM plan and pitch, and one real GTM action shipped (posted, texted, or shared).
Office hours (strongly recommended)
- Friction pass — find and fix your 3 worst friction moments
- Demo rehearsal — pair up, run your 60-second demo timed, twice
- Final push — all code to GitHub, live URL works, demo path locked
Optional office hours: 3:00-4:00 PM ET / 12:00-1:00 PM PT for polish, demo rehearsal, GitHub push, and deployment.
Day 8 (Sunday) — Demo Day
Theme: "The thing you came here to do."
Schedule (2:00-4:00 PM ET / 11:00 AM-1:00 PM PT)
- 2:00-2:10 — Welcome, demo order, and quick mic/screen-share check
- 2:10-3:40 — Live student demos (parents invited): 60 seconds to demo + 30 seconds for Q&A / feedback
- 3:40-4:00 — Awards + closing (Pixel Perfect, Under the Hood, Built for Humans)
Outcomes — every student leaves with:
- A working AI-powered tool deployed on Replit
- A working automation — a scheduled job, bot, or recurring task that runs on its own
- A 60-second demo video
- A GitHub portfolio repo + live Replit URL; public only when parent-approved
- A spot in the alumni community
- AI Builders Camp swag
What everyone leaves with — and where bonus moves go
Every student finishes the same daily goal. Bonus moves are opt-in — kids who finish early or want to push further pick them up. We don't expect every student to hit every bonus, and that's fine. The goal is a tool that works, not a checklist.
| Everyone | Bonus moves | |
|---|---|---|
| Editor | Replit | + Replit agent for scaffolding |
| UI | Hand-built in Replit | Custom component design |
| Persistence | localStorage / Replit DB | + form data, saved state |
| Automation | One working automation (scheduled job, bot, or recurring task) | Interactive bot via Replit Connector |
| Publishing | Replit deployment + GitHub repo + screen-share demo | Public launch with parent approval |
| Final scope | One feature, polished | Full workflow + parent-approved public launch |
Why this curriculum, not "intro to Python"
Most kid coding camps teach syntax. This camp teaches building — using AI as a multiplier, not a crutch. The students who win the next decade are the ones who can hold a product in their head, scope it, ship it, defend it, and iterate. Students don't leave knowing every tool. They leave knowing how to turn an idea into a working product they can actually show someone. In two weeks, they go from “I wonder if I could build that” to “here, look — I made this.” That's the whole point.