AI Power User → Builder
Learn AI by moving from use to control.
AI Lab is the Turtleand curriculum spine: a staged path from practical AI workflows to agent architecture, retrieval systems, and production judgment.
Hermes
Voice-first agent operations, scheduling, tools, skills, and reliable human-AI workflows.
Enter →AI Atlas
Interactive map of AI tools I've charted on my journey.
Explore →Handbook
Human-AI collaboration principles. Quick reference for teams and individuals.
Enter →Start here
New to AI? Use AI for Beginners.
AI for Beginners is the zero-start doorway into AI Lab. It explains the basic mental model, keeps human judgment at the center, and helps you choose whether to stay in the curriculum or branch toward Portal, AI Atlas, Build, or Hermes.
Open AI for Beginners →Curriculum spine
Program roadmap
Open a module to see the next learning artifacts. The active module isModule 2: AI Integration & Orchestration.
Module 0Fast Track Setup
- AI for Beginners
A zero-start doorway into AI Lab, with human judgment at the center.
- Running inference locally
Easily setup & run an LLM on your own machine.
- Agent notifications
Get notified when your agent finishes using hooks and ntfy.
- Access & secrets
Configure API keys and secrets securely for model access.
- Safety baseline
Practical safety for AI infrastructure — firewalls, audits, and access control.
- Visit module page
Full overview, resources, and artifacts for this module.
Module 1AI Power User
- Multi-tool AI workflows
Running 3+ AI tools in parallel, choosing the right tool for each task.
- Prompt mastery
Beyond basics — structured prompting for autonomous systems.
- Model selection & economics
When to use Opus vs Sonnet, quota management, and cost-per-output thinking.
- Voice & multimodal workflows
Voice-to-AI pipelines, TTS output, and audio as a primary instruction method.
- Human-AI collaboration handbook
A quick-reference guide to principles for human-AI collaboration, from mindset to workflow.
- Visit module page
Full overview, resources, and artifacts for this module.
Module 2AI Integration & Orchestration
- From OpenClaw to Hermes
Why persistent-agent experiments graduated into a leaner, human-centered Hermes operating layer.
- Automation pipelines
Cron-driven AI workflows, multi-stage processing, and overnight analysis.
- AI-first lifestyle
Morning briefings, memory systems, and making AI work while you sleep.
- Content & distribution
AI-augmented content creation, cross-platform publishing, analytics.
- Building a knowledge-base MCP server
I built an MCP server that exposes 51 articles to any AI client. Two tools, zero config, no API keys. Here's how it works.
- Living institutional memory
How organizations can use AI to ensure no knowledge ever degrades and no lesson is ever relearned.
- Visit module page
Full overview, resources, and artifacts for this module.
Module 3Builder Foundations
- LLM interfaces
Client libraries, request/response patterns, streaming.
- Inputs & outputs
Structured prompts, JSON/tool outputs, function calling.
- Embeddings & vector basics
Understanding embeddings, similarity search, when you need vectors.
- Context is not memory
Why agents need curated memory, not just bigger context windows.
- Planning and decomposition as control structures
How plans keep agent work bounded, observable, and adaptable.
- Prompting & reasoning patterns
Advanced prompting, CoT, evaluation, micro-tools.
- Visit module page
Full overview, resources, and artifacts for this module.
Module 4RAG & Retrieval SystemsPlanned
- Vector databases
ChromaDB, pgvector, Pinecone — setup and comparison.
- Document indexing & chunking
Processing your own content for semantic search.
- RAG pipeline design
End-to-end retrieval-augmented generation.
- Observability
LangSmith, Langfuse — tracing, evaluation, cost tracking.
- Visit module page
Full overview, resources, and artifacts for this module.
Module 5Agent Architecture
- Agentic AI systems learning path
A concise learning map for agents: loop, tools, memory, planning, approval, reliability, trust infrastructure, routing, and coordination.
- AI Disruption Beyond Horseless Carriage Thinking
Seven AI-native shifts that move beyond treating AI as a better old tool: operators, agent-accessible environments, code steering, skills, agent services, adaptive learning, and compound systems.
- Agent frameworks
LangGraph, CrewAI — multi-agent orchestration.
- MCP protocol
Model Context Protocol — building tool servers for AI.
- Agent loop & control flow
A beginner lesson on how useful AI agents observe state, choose bounded actions, inspect results, and keep humans responsible for risky steps.
- Agent tool use as a controlled action cycle
A beginner lesson on how tools give agents hands and senses, why tool calls need permission boundaries, and how results should be checked before the next action.
- Human Agency Is Not a Button
A beginner lesson on why responsible agent systems need authority boundaries, not just approval prompts.
- Smart Is Not Reliable
Why agent reliability is a system property built from checks, constraints, recovery paths, observability, and human escalation.
- Trust Is Earned Through Evidence, Not Vibes
How agent systems earn trust through scoped tasks, permission boundaries, observability, evaluations, fallback paths, and independent verification.
- Model Choice Is System Design
How agent systems choose models by task difficulty, risk, cost, latency, privacy, verification needs, and human authority boundaries.
- LLM-to-LLM Interaction Patterns
How same-provider, cross-provider, and two-model LLM-to-LLM systems work when they are bounded by roles, turns, evidence, and a responsible orchestrator.
- More Agents, More Coordination
Why multi-agent systems trade individual workload for delegation, evidence, conflict resolution, and synthesis work.
- Visit module page
Full overview, resources, and artifacts for this module.
Module 6Production AI SystemsPlanned
- Local model serving
Ollama, vLLM — cost optimization, privacy, fine-tuning experiments.
- Packaging & deployment
CLI tools, web UIs (Streamlit/FastAPI), Docker.
- Evaluation frameworks
Systematic prompt testing, A/B evaluation, quality scoring.
- Shipping AI products
From personal infrastructure to products others use.
- Visit module page
Full overview, resources, and artifacts for this module.