The domains 2nth brings AI to — and the technology foundations that make it run. Every skill node is context you can read, your agent can load, and your team can operate with.
Every skill node in the tree serves humans and agents simultaneously — the same knowledge, consumed three different ways.
Ten skill nodes aren't ten times more useful than one — they're a hundred times more useful. Every node connects to every other node, and your agent can traverse the entire tree.
Every piece of content serves the human learning and the agent operating — simultaneously.
Explainers built by practitioners, not content farms. Honest tradeoffs, real architecture decisions, and practical guidance.
Load any branch of the skills tree into your agent's context. Transform a generic model into a domain-aware operator.
Canonical 2nth skill domains — the structure every 2nth agent, partner, and build is grounded in. Each card is a branch of the tree, with production skills underneath and more landing every month.
The full agentic stack — seven orchestration frameworks (ADK, LangGraph, CrewAI, AutoGen, Anthropic SDK, OpenAI SDK, AgentForce), three interop protocols (A2A, MCP, Agent Skills), six models (Claude, GPT, Gemini, Gemma, Llama, Mistral / Qwen), and four inference platforms (Ollama, vLLM, LiteLLM, OpenRouter). The orchestration layer over data and systems-of-record. Where the rest of the tree turns into work.
Deepest branchCloudflare, Google, Microsoft, MuleSoft, modern web stacks, Frappe, runtime tooling. The infrastructure and integration layer every other domain runs on.
Production · 9 leavesCRM, ERP, ECM, HR, operations. The systems businesses actually run on — integrated, API-first, and built to carry AI inside the workflow, not bolted on.
Live · 6 leavesAnalytics, warehousing, engineering, the CDP / unified-profile layer behind AgentForce. The layer that turns raw events into decisions — and the one regulators care most about.
Capital markets, modelling, reporting, accounting, treasury, tax. The practices that move money and risk — with South African FSCA and SARB context where it matters.
Commercial, disputes, property, employment, IP, corporate, data privacy, fundraising. Contract intelligence and regulatory practice — the domains where getting it wrong isn't an option.
Clinical systems, operations, finance, digital, public health, pharmacy, mental health. The practice areas where the stakes are patient outcomes, not just revenue.
Hardware, firmware, connectivity, sensors, power, platform. The skills for building systems that live outside the cloud — and survive load shedding doing it.
Curriculum design, assessment, higher ed, K12, corporate training. The infrastructure behind learning that compounds — across CAPS, IEB, Cambridge, IB.
Coaching, leadership, teams, organisational psychology. The discipline of growing humans — frameworks like the Enneagram alongside coaching practice, facilitation, and the development conversation that compounds across a career.
LiveTokens, components, frameworks, motion, typography, accessibility, Figma, AI design tools, Open Design. The interface layer — twin craft to architecture, opinion-heavy and load-bearing on every product surface.
LiveopenBIM, IFC, AI takeoff, 4D programme, materials and quality, RFIs, SA compliance. The least digitalised major industry on earth, addressed with open standards and the Gridline Properties × 2nth.ai openBUILD AI partnership.
Recently published skill nodes — each one making the tree more connected and every agent more capable.
Cross-page audit correcting three overstated SA-residency claims: Claude on Bedrock af-south-1 (endpoint + logs only, inference routes globally), Azure OpenAI in southafricanorth (Regional PTU only, not Global Standard), and the "only major LLM vendor" framing (Vertex AI africa-south1 also present). Memo with primary sources.
Salesforce's agent platform. Atlas Reasoning Engine, five-attribute agents, Flex Credits at $0.10. Grounded in Data Cloud, wrapped by the Einstein Trust Layer.
The enterprise CRM platform — SObject + SOQL + Apex + Flow on Lightning, with AgentForce, Data Cloud, Einstein, MuleSoft, Slack, and Tableau wrapped around it.
The Customer Data Platform behind AgentForce. Real-time ingestion, identity resolution, harmonisation across structured + unstructured sources, RAG-ready semantic search.
Anypoint Platform plus the AgentForce-era overlay — API Catalog, MuleSoft MCP Server, Agent Fabric. The path that turns existing enterprise APIs into MCP servers.
Salesforce's BI platform. Tableau Cloud / Pulse / Next. Zero-copy with Data Cloud; MCP-callable from AgentForce via Tableau Next.
Anthropic-originated open standard for packaging agent capabilities as folders. SKILL.md + progressive disclosure. Native in Claude Code, Claude.ai, and the API.
Start free. Go deeper as the value compounds for your team and your agents.
Technology explainers and framework overviews. Read, learn, and load into your agent's context.
Operational skill nodes, context bundles, deployment templates, and curated learning paths.
Your IP operationalized as a skills tree. Private branches, co-branded nodes, and agent-ready context for your domain.