Hands-on, copy-paste-ready training guides for Backboard.io — the unified API for AI assistants, memory, documents, and multi-model routing.
Note
Docs: docs.backboard.io · Dashboard: app.backboard.io
- A Backboard account with an API key (get one here)
- Python 3.7+ or Node.js 18+ (depending on the codelab)
- Install the SDK:
# Python
pip install backboard-sdk
# JavaScript / TypeScript
npm install backboard-sdk| # | CodeLab | Duration | Language | What You'll Build |
|---|---|---|---|---|
| 01 | Hello Backboard | ~5 min | Python | Your first AI message using the one-call API |
| 02 | Hackathon Speedrun | ~10 min | Python | Stateful chat + document RAG in under 50 lines |
| 03 | Solo AI Stack | ~20 min | Python | Unified memory + document search in a single API |
| # | CodeLab | Duration | Language | What You'll Build |
|---|---|---|---|---|
| 04 | Multi-Provider Orchestration | ~25 min | Python | Automatic failover across OpenAI, Anthropic, Google |
| 05 | Benchmarking Scorecard | ~20 min | Python | Automated latency, cost, and accuracy comparisons |
| # | CodeLab | Duration | Language | What You'll Build |
|---|---|---|---|---|
| 06 | Multi-Tenant Isolation | ~25 min | TypeScript | Per-user AI sessions with cross-thread memory |
| 06b | Production Session Broker | ~20 min | TypeScript | Hardened session broker with model failover |
| 07 | Enterprise AI Gateway | ~30 min | TypeScript | OpenAI-compatible proxy with telemetry & model swapping |
| 07b | Enterprise Governance (Python) | ~30 min | Python | PII redaction & department-level compliance gateway |
| # | CodeLab | Duration | Language | What You'll Build |
|---|---|---|---|---|
| 08 | Edge & Mobile | ~20 min | TypeScript | Zero-dependency client for constrained devices |
| 09 | EdTech Adaptive Tutor | ~20 min | Python | AI tutor with cross-session student progress tracking |
Important
R-CLI is Backboard's AI coding agent for the terminal. Give it a goal, and it uses 13 built-in tools to read code, edit files, run commands, and iterate until the job is done. Request beta access →
| # | CodeLab | Duration | Language | What You'll Build |
|---|---|---|---|---|
| 10 | Hello R-CLI | ~5 min | Interactive | Your first R-CLI session — models, tools, cost & context |
| 11 | Recursive Debugging | ~15 min | Interactive | Let R-CLI fix failing tests autonomously |
| 12 | Code Review & Security Audits | ~20 min | Interactive | AI-powered code review and security scanning |
| 13 | Project Memory & Skills | ~20 min | Interactive | Persistent workspace knowledge, semantic indexing, and reusable skills |
| 14 | Expert Mode & Multi-Model | ~20 min | Interactive | Two-model architecture, model switching, and cost optimization |
| 15 | Browser & Desktop Automation | ~20 min | Interactive | Browser harness, computer tool, and end-to-end test flows |
| 16 | CI/CD Integration | ~20 min | YAML/Bash | One-shot mode for GitHub Actions, pre-commit hooks, and pipelines |
| 17 | MCP Server Ecosystem | ~20 min | Interactive | Adding MCP servers, combining external tools with built-in ones |
| 18 | Semantic Code Indexing | ~15 min | Interactive | Vector and text-only codebase search for conceptual queries |
Tip
Build a complete AI-powered research platform inside Google Sheets — progressively. Each lab adds one Backboard feature. By Lab 25, you have a multi-agent, memory-native, document-grounded AI product with a full settings UI, usage analytics, and scheduled automation.
| # | CodeLab | Duration | Language | What You'll Build |
|---|---|---|---|---|
| 19 | Hello Sheets AI | ~10 min | Apps Script | Your first AI message from a spreadsheet with web search |
| 20 | Model Switcher | ~15 min | Apps Script | Dynamic model/provider selection with settings sidebar |
| 21 | Persona Engine | ~20 min | Apps Script | Configurable AI assistants with preset personas |
| 22 | Memory Intelligence | ~25 min | Apps Script | Persistent cross-run memory as semantic cache |
| 23 | Document RAG | ~20 min | Apps Script | Strategy-grounded analysis with uploaded documents |
| 24 | Multi-Agent | ~25 min | Apps Script | Parallel AI workers with specialized personas |
| 25 | Command Center | ~25 min | Apps Script | Complete product: settings, analytics, automation |
Important
Master R-CLI's workspace automation: scaffold projects, generate agentic code, configure memory and MCP servers, and deploy production-ready AI agents — all from your terminal.
| # | CodeLab | Duration | Language | What You'll Build |
|---|---|---|---|---|
| 26 | Workspace Setup | ~15 min | Interactive | Scaffold a project workspace with BACKBOARD.md, git, and environment config |
| 27 | Generate Agent | ~20 min | Interactive | Use R-CLI to generate a full AI agent codebase from natural language |
| 28 | Memory & MCP Config | ~20 min | Interactive | Configure persistent memory, semantic indexing, and MCP server integrations |
| 29 | Test & Debug Agent | ~20 min | Interactive | Validate, debug, and iterate on your generated agent using R-CLI's agentic loop |
| 30 | Deploy & Productionize | ~25 min | Interactive | Production deployment: Docker, CI/CD, environment configs, and monitoring |
Tip
Build a complete AI-powered travel app from an empty folder — one R-CLI prompt per lab. Each lab provides a short prompt and a detailed spec prompt, showing how prompt quality scales output quality. By Lab 35, you have a production Next.js travel app with an AI concierge powered by Backboard's one-call API.
| # | CodeLab | Duration | Language | What R-CLI Will Build |
|---|---|---|---|---|
| 31 | Project Scaffold & Specs | ~15 min | Interactive | Next.js project, spec docs, Wikipedia API client, 30 Italian cities |
| 32 | Frontend: City Grid & Detail | ~20 min | Interactive | Responsive card grid, search, filters, favourites, city detail pages |
| 33 | Backend: Prisma & API Routes | ~20 min | Interactive | SQLite database, Prisma schema, REST APIs, Zod validation, hook migration |
| 34 | Trip Planner & Map View | ~20 min | Interactive | Drag-and-drop itinerary, Leaflet map with markers, print-friendly CSS |
| 35 | AI Travel Agent — Backboard API | ~25 min | Interactive | Conversational AI agent: RAG, thread memory, tool calls — one API endpoint |
- Pick a track that matches your experience level and use case
- Open the CodeLab markdown file — each one is self-contained
- Follow step-by-step — copy-paste the code blocks, they run without modification
- Check the validation section at the end of each lab to verify your output
📌 CodeLab Structure
Every CodeLab follows a consistent structure:
| Section | Description |
|---|---|
| 🎯 Overview & Objectives | What you'll build |
| 🔧 Prerequisites & Setup | Environment configuration |
| 🛠️ Step-by-Step Implementation | Modular, commented code blocks |
| ✅ Validation & Verification | Expected outputs and checklists |
| 🔍 Under the Hood | How the architecture works |
| ➡️ Next Steps & Challenges | Extend your build |
| Resource | Link |
|---|---|
| 📚 Documentation | docs.backboard.io |
| 🔌 API Reference | docs.backboard.io/api-reference |
| 🖥️ Dashboard | app.backboard.io |
| 🏗️ Architecture Overview | docs.backboard.io/concepts/architecture |
| ⚡ SDK Quickstart | docs.backboard.io/sdk/quickstart |
Found an issue or want to suggest an improvement? Open an issue or submit a pull request.