Pages tagged "architecture"

5 pages tagged with architecture. ← all tags

• The Staged-Compiler Pattern: Chaining AI Skills from Strategy to Production An architecture pattern for AI work pipelines — two compilers joined by a frozen JSON contract, human gates at irreversible steps, parallel fan-out with central allocation.
• Advisor Strategy — Pairing a Smarter Model as an Occasional Advisor With a Cheaper Executor Anthropic's advisor pattern (April 2026): the executor model (Sonnet or Haiku) handles tasks end-to-end while consulting an advisor model (Opus) only on hard decisions. Server-side, single API request. Sonnet+Opus advisor: +2.7pp on SWE-bench at -11.9% cost. Haiku+Opus: 41.2% on BrowseComp vs 19.7% solo, 85% cheaper than Sonnet alone.
• Context Engineering — Designing Information Flow for AI Agents How to structure tool responses so AI agents can reason effectively across multiple calls
• Claude Managed Agents vs. DIY Agent Infrastructure When to use Anthropic's managed platform vs. building your own agent orchestration with Messages API
• Multi-Agent Patterns — When Two Agents Beat One Practical patterns for combining AI agents: dispatcher + deep worker, content pipelines, and self-learning systems