Research by Ry Walker

A comparison of 10 agent coordination protocols, layers, and frameworks — from the A2A standard (v1.0 under the Linux Foundation) and the layered protocol stack (AG-UI, Zed ACP, AGNTCY) to orchestration frameworks (CrewAI, AutoGen, LangGraph). Who coordinates agents, and how?

Comparative analysis of 13 AI agent frameworks: Agno, AutoGen, AWS Strands Agents, Cargo AI, CrewAI, Google ADK, LangChain/LangGraph, LlamaIndex, Mastra, Microsoft Agent Framework, OpenAI Agents SDK, Pydantic AI, and Vercel AI SDK — covering architecture, observability, and use cases.

A comparison of 10 tools for agent memory and self-improvement — from memory layers like Mem0, Letta, Hindsight, and Cognee to evolving playbook systems like ACE, plus the research foundations driving the category.

A comparison of 42 coding agent desktop and terminal apps — including Cursor, Claude for Mac, Codex App, Warp Oz, Emdash, OpenCode, Sculptor, cmux, Paseo, and 30+ more. Every member re-verified June 2026, including the category's first deaths.

A comparison of 20 agentic skills frameworks and tools — from methodology enforcers like Superpowers and BMAD to official catalogs and marketplaces from Anthropic, OpenAI, Google, and Vercel, plus orchestration platforms, security-first registries, and the SKILL.md standard.

A comparison of 13 leading voice AI platforms for building agentic voice applications — OpenAI Realtime, Gemini Live, AWS Nova 2 Sonic, ElevenLabs, Cartesia, Gradium, Rime, NVIDIA PersonaPlex, Vapi, Retell AI, LiveKit Agents, Pipecat, and Deepgram Aura.

Category analysis of 7 platforms that organize AI agents as employees in company structures — from Paperclip's 70k-star org chart orchestration to Edict's Tang Dynasty governance model. Covers the category's consolidation: who's growing, who's dormant, and who got reclassified.

A comparison of 19 leading sandbox and infrastructure platforms for AI agents — E2B, Daytona, Modal, Sprites, Vercel Sandbox, Cloudflare Sandbox SDK, AWS AgentCore Code Interpreter, Google Agent Sandbox, NVIDIA OpenShell, Blaxel, Northflank, OpenSandbox, Runloop, CodeSandbox SDK, Microsandbox, AIO Sandbox, ComputeSDK, Quilt, and Zeroboot.

Comprehensive comparison of 14 AI coding assistants — Aider, Amp, Claude Code, Cline, Continue, Crush, Goose, Kilo Code, OpenCode, OpenHands, Refact.ai, and Sweep, plus the departed Roo Code and open-source background agents. Features, pricing, and ideal use cases.

A comparison of 11 tools that track AI coding impact — attribution, token costs, productivity metrics, and code quality — across the new wave of engineering intelligence platforms built for the AI-assisted development era.

A comparison of 13 leading AI inference platforms — Baseten, Together AI, Fireworks AI, Modal, Replicate, Groq, DeepInfra, Cerebras, SambaNova, Vercel AI Gateway, Fal, OpenRouter, and Nebius.

Category analysis of 6 AI QA agents and agentic testing platforms — Momentic, QA Wolf, Ranger, Spur, Expect, and the departed Octomind. From autonomous browser-testing suites to tools that verify coding-agent output.

Category analysis of 11 AI-native sales and GTM platforms replacing legacy CRM and outbound tools. Covers Rox, Monaco, Aurasell, SuperAGI, Artisan, AiSDR, Landbase, Alta, Salesforge, 1mind, and the departed Persana — from agentic CRM to AI SDR agents to live-call AI sellers.

A comparison of 13 autonomous coding agents, orchestrators, and collaboration platforms — AgentHub, Agent Orchestrator, Gas City, Gastown, Cosine (Genie), GPT Engineer, Metaswarm, oh-my-claudecode, Optio, Pythagora, Ralph, Smol Developer, and Symphony — that aim to automate software development with minimal human intervention.

Category analysis of 16 autoresearch tools — autonomous AI agents that run experiment loops, deep web research, and scientific discovery. Covers Karpathy's autoresearch, Kosmos, Autoscience, Google Co-Scientist, AutoResearchClaw, GPT Researcher, Tongyi DeepResearch, AI Scientist, and more.

A comparison of 14 cloud coding agent platforms — Tembo, Devin, Factory, Ona, Replicas, Niteshift, Claude Code on the web, Codex cloud, and more. Agents run in managed cloud environments, take delegated tasks from Slack, Linear, or tickets, and return pull requests.

Category analysis of 14 code intelligence tools that give AI coding agents structural understanding of codebases. Covers CodeGraph, GitNexus, Serena, claude-context, grepai, Augment Context Engine, Repomix, and more — from local-first knowledge graphs to semantic search and context packing.

A comparison of 9 tools that run containerized workloads with VM-level isolation: Firecracker, gVisor, Kata Containers, Cloud Hypervisor, Apple Containerization, libkrun, Edera, Hypeman, and Unikraft. The "containers aren't isolated enough" problem, solved.

Category analysis of 10 developer trust and supply chain security tools. Covers Vouch (contributor trust), git-ai (AI code provenance), Sigstore (artifact signing), SLSA (build provenance), in-toto (attestation), zizmor (CI auditing), Socket.dev (proactive detection), OpenSSF Scorecard (project scoring), and GUAC (knowledge graph).

A comparison of 8 official coding agents from AI foundation labs — Claude Code (Anthropic), Codex (OpenAI), Gemini CLI and its Antigravity CLI successor (Google), Grok Build (xAI), Qwen Code (Alibaba), Kimi Code (Moonshot), and Mistral Vibe — the only agents built by the model makers themselves.

Analysis of 15 in-house coding agents: how Stripe, Google, Meta, OpenAI, Harvey, Uber, Spotify, Shopify, Coinbase, Cloudflare, Browserbase, and other tech companies are building custom agents instead of buying off-the-shelf.

A comparison of 8 local sandboxing tools for AI coding agents — anthropic-sandbox-runtime, nono, landrun, fence, yolobox, litterbox, yolo-cage, and shai — plus the first-party sandboxes now shipping inside Codex and Claude Code. Protect your machine while letting agents run wild.

A comparison of 7 MCP integration platforms — Arcade, Composio, Klavis AI, Nango, Paragon, Pipedream, and Zapier MCP. How AI agents get authenticated access to third-party SaaS tools via managed MCP servers, tool catalogs, and managed OAuth.

Comparative analysis of 12 minimalist markdown note-taking apps for Mac — from quick capture to long-form writing.

Comparative analysis of 46 platforms building personal AI agents: from managed solutions like Manus, Lindy, Gemini Spark, Microsoft Scout, Perplexity PC, and Poke to self-hosted alternatives including OpenClaw, Hermes Agent, Pi, NanoBot, AstrBot, ZeroClaw, QwenPaw, and the full -claw ecosystem — plus NVIDIA NemoClaw for enterprise.

Comparative analysis of 8 AI product management agents: BrainGrid, BuildBetter, ChatPRD, Korey, Modem, Nalvin, Product Forge, and Product Manager Skills. From spec generation to feedback triage to meeting intelligence.

Comparison of 11 SEO and GEO agent skills for Claude Code and AI coding agents — from GEO-first audit tools and deterministic rule engines to full marketing skill libraries. Covers citability scoring, AI crawler analysis, live citation checking, content optimization, and agency workflows.

Comparative analysis of 14 platforms for deploying shared AI agents across teams — covering product planning, data analysis, customer support, and operations. Coding-first platforms are covered separately.