> kb/ai-stack-2026.md · STACK · 12 MIN

FIELD MAP

THE AI STACK — MAY 2026

The application layer most teams ship on is now ten distinct layers deep, wrapped by two rails that touch every one of them. This is a working map, not a buyer's guide: where each category sits, what a few representative providers do, and how the pieces connect.

Read it top to bottom — surface to silicon. The left rail, Observability, and the right rail, Governance, are not steps in the flow; they are concerns that cut across all ten layers. Tap any provider in the diagram to jump to its explanation and an outbound link below.

THE AI STACK — MAY 2026

Tap any item for details ↓

01End-User Surfaces

Cursor

AI-first code editor; agentic edits and codebase-wide changes from natural language.

Visit provider

Perplexity

Answer engine: conversational search with live sources and citations.

Visit provider

Claude

Anthropic's assistant across web, desktop and mobile, tuned for long-context work.

Visit provider

02Agent Runtimes

Claude Code

Terminal-native agentic coding from Anthropic; delegates multi-step engineering tasks.

Visit provider

Devin

Cognition's autonomous software engineer that plans and executes end-to-end.

Visit provider

Replit Agent

Builds and deploys full apps from a prompt inside Replit's cloud IDE.

Visit provider

Codex

OpenAI's coding agent for the cloud and CLI, running tasks in isolated sandboxes.

Visit provider

Cursor Agent

Cursor's background agent mode for parallel, longer-running coding work.

Visit provider

03Orchestration Frameworks

LangGraph

Graph-based orchestration for stateful, multi-step agent workflows (LangChain).

Visit provider

Microsoft Agent Framework

Microsoft's unified agent framework, consolidating Semantic Kernel and AutoGen.

Visit provider

Google ADK

Google's open-source Agent Development Kit (Python, Java, Go, TypeScript).

Visit provider

04Protocol Layer

MCP

Model Context Protocol (Anthropic): a standard way to connect models to tools and data.

Visit provider

A2A

Agent2Agent: cross-vendor agent interoperability; created by Google, now Linux Foundation.

Visit provider

AG-UI

Agent-User Interaction protocol (CopilotKit): event stream between agent backends and frontends.

Visit provider

05Memory

Mem0

Drop-in memory API combining vector, graph and key-value stores for personalization.

Visit provider

Letta

OS-style agent memory with paging between context and archival storage (formerly MemGPT).

Visit provider

Zep

Temporal knowledge-graph memory (Graphiti) that tracks how facts change over time.

Visit provider

06Retrieval

Cohere Rerank

Reranking models that reorder candidate passages by true relevance.

Visit provider

07Storage

pgvector

Postgres extension adding vector similarity search to an existing database.

Visit provider

Turbopuffer

Serverless vector and full-text search built on object storage for low cost at scale.

Visit provider

08Model Gateway

Portkey

AI gateway adding routing, caching, guardrails and observability across providers.

Visit provider

LiteLLM

Unified SDK and proxy exposing 100+ model providers behind one OpenAI-style API.

Visit provider

09Foundation Models

Claude (Anthropic)

Anthropic's Claude model family, tuned for reasoning, coding and long context.

Visit provider

10Inference + Compute

AMD MI400

AMD's Instinct MI400-series AI accelerators — AMD's datacenter challenge to NVIDIA.

Visit provider

Google TPU

Google's Tensor Processing Units for training and serving on Google Cloud.

Visit provider
MEMBER · FREE

Read the full article or download as PDF

The full article and the PDF are member content. Magic-link login, no credit card, no risk — both available immediately.