Anthropic ecosystem
Where we sit, what we ship on, who we answer to
Adaptation AI is built on Anthropic's Claude, Anthropic's framework, and Australian compliance discipline. Below: the stack we run, the curriculum we teach from, the regulatory posture you'll inherit, and the partner status as it stands today.
Stack we ship on
One vendor for the model, one protocol for the integrations, one SDK for the agents. Claude-exclusive isn’t a hedge — it’s the position. Every primitive below is in production on real client work.
Claude Opus 4.7
Anthropic's most capable model. Opus 4.7 (released Feb 2026, refreshed Apr 16) introduces adaptive thinking, the new xhigh effort tier, task-level budget controls, high-res vision (3.75MP), and a new tokenizer. 1M token context (GA March 13, 2026), 128K output. Priced at $5/$25 per MTok. Ideal for complex analysis, long-horizon agentic tasks, and sustained multi-step workflows.
Claude Sonnet 4.6
High-performance model with 1M token context (GA at standard pricing since March 13, 2026) and 64K output tokens. Priced at $3/$15 per MTok. Major improvements in coding, computer use, and instruction following. The optimal balance of intelligence and speed for production workloads.
Claude Haiku 4.5
Ultra-fast and cost-efficient. Best for high-volume tasks like classification, extraction, and real-time user interactions.
Model Context Protocol (MCP)
The de facto industry standard for connecting AI to external data sources and tools. Now under Linux Foundation governance with 97M+ monthly SDK downloads and MCP v2 beta.3 bringing multi-agent communication.
Agent SDK
The framework for building autonomous AI agents with built-in guardrails, tool use, and multi-agent orchestration.
Claude Code
Agentic coding tool with voice mode, /loop scheduling, and 1M token context. 2M+ VS Code installs, JetBrains beta plugin. Operates directly in your terminal, understanding full codebases and executing complex development tasks.
Claude Cowork
Desktop agent for autonomous knowledge work (research preview). Reads, edits, and creates files locally, breaks complex tasks into parallel subtasks, and integrates with enterprise tools via plugins and MCP connectors.
Claude Code Security
Vulnerability scanner using Claude Opus 4.7 to reason about code contextually — tracing data flows, mapping component interactions, and finding business logic flaws traditional tools miss.
Agent Teams
Multi-agent coordination where Claude Code instances work as a team — one as lead, others as specialists. Teammates share findings and converge on solutions for parallel development.
Skills & Plugins
A rapidly growing ecosystem of hundreds of plugins and thousands of skills with the CCPI package manager. Skills auto-activate based on context, Plugins bundle MCP servers and agents into shareable packages.
Adaptive Thinking
Claude's advanced reasoning capability that dynamically adjusts thinking depth based on task complexity. From quick classifications to deep multi-step analysis, ensuring optimal quality-latency tradeoffs.
Where the data actually sits
Diligence-axis essentials. Australian sovereignty isn’t a marketing line — it’s a regulatory posture. Pick the cloud your governance already approves; we deploy Claude inside it.
Amazon Web Services
Sydney Region (ap-southeast-2) — Claude via Amazon Bedrock
Google Cloud
Australia Region — Claude via Vertex AI
Microsoft Azure AI Foundry
Deploy Claude through Azure's enterprise infrastructure with native identity management and security controls.
Anthropic opened its Sydney office in March 2026. The Australian ecosystem is one year ahead of where it was, and a year behind where it’ll be — which is the right window for a boutique that already speaks the framework.
Curriculum we teach from
Anthropic Academy is the public catalogue. The AI Fluency track in the first column is the framework we anchor every engagement to — Automation, Augmentation, Agency across the four Ds. We use the framework under fair attribution to Dakan, Feller, and Anthropic (CC BY-NC-SA 4.0); productisation is ours.
AI Fluency (the spine)
- Framework & Foundations — A/A/A x 4D
- Delegation, Description, Discernment, Diligence
- AI Fluency for educators / students / nonprofits
- Teaching AI Fluency (train-the-trainer)
Product training
- Claude for Business
- Claude Enterprise Administration
- Claude API Fundamentals
- Claude for Teams
Developer deep-dives
- Advanced prompt engineering
- Model Context Protocol (MCP)
- Agent SDK & agentic architectures
- Claude Code for development teams
- Constitutional AI & safety alignment
Compliance roadmap — Diligence in dates
Diligence is the fourth D and the first thing procurement asks about. Here’s where we are, and where we’re headed — without promising a posture we don’t hold yet.
Certification status — what we’ve actually achieved
Two Anthropic credentials matter for buyers checking us out: Partner Network membership and Claude Certified Architect. We’re honest about where we are. Both are in flight. The dates aren’t hidden.
Anthropic Partner Network
Claude Certified Architect
Want to test the framework on a real workflow?
Free scope, free build, you only pay if it works. 30-min call books straight to a real engineer.