Section I
Personal & Always-On Agents
Self-hosted agents that run 24/7, connect to your messaging apps, and act on your behalf. This category was created by OpenClaw in late 2025 and has spawned an entire ecosystem in under six months.
The originals — defined the category
Peter Steinberger → OpenAI / Foundation
The original viral personal agent. Gateway architecture routes all messages through a central controller. 5,700+ community skills on ClawHub. Model-agnostic, 50+ integrations. Now under independent foundation with OpenAI backing after Steinberger joined OpenAI Feb 2026.
Nous Research
Agent-loop architecture. Persistent memory, autonomous skill creation, self-improving learning loop, user modelling via Honcho. Positioned between Claude Code CLI and OpenClaw. Built-in hermes claw migrate command for OpenClaw users.
The security response — built because OpenClaw's CVEs were alarming
Gavriel Cohen / Qwibit
~700 lines of TypeScript — auditable in 8 minutes. Every agent runs in an isolated Linux container that self-destructs after each message. Built on Claude Agent SDK. Container-per-chat-group isolation prevents data leakage between contexts.
NEAR AI
Rebuilt from scratch in Rust. WASM sandboxing with cryptographic capability tokens per skill. TEE-backed execution, encrypted credential vaults, zero telemetry. PostgreSQL with AES-256-GCM. The only personal agent suitable for regulated industries.
NVIDIA
Not a standalone agent — a security wrapper over OpenClaw. Adds OpenShell kernel-level sandbox via Linux security modules. Policy-as-YAML. Ships with NVIDIA Nemotron open models. Announced by Jensen Huang at GTC March 2026.
The minimalists — lighter, hackable, hardware-friendly
Community
Runtime rewritten in Rust. 3.4MB single binary. Zero cloud dependencies, entirely local. Ultra-low memory footprint for users who want complete data sovereignty.
Community
4,000 lines of readable Python. Runs on Raspberry Pi. Supports local models via Ollama. No plugin system, no config sprawl. Built for developers who want to understand every line of their agent.
Community
Runs on $10 boards. Under 10MB RAM. Designed for constrained hardware — Raspberry Pi Zero, embedded systems. Single-purpose, maximum portability.
Multi-agent and enterprise variants
Community
Collaborative multi-agent OS built on Matrix rooms. Transparent, human-in-the-loop task coordination. Supports OpenClaw, NanoClaw, and ZeroClaw agents as workers. MinIO shared filesystem for inter-agent exchange.
Eigent
Desktop multi-agent workforce. Connects to your context, controls browser and desktop apps to automate real work. Visual interface rather than messaging-native. Non-developer accessible.
Anthropic
Not a full agent framework. Lets developers control local Claude Code sessions from Telegram and Discord. Specifically for developer workflows, not general personal assistant use. Launched March 2026.
Head-to-head comparison
| OpenClaw | Hermes | NanoClaw | IronClaw | NemoClaw | ZeroClaw | Nanobot | |
|---|---|---|---|---|---|---|---|
| Architecture | |||||||
| Core design | Gateway / controller | Agent loop / learning | Container-per-agent | WASM + Rust | Wrapper over OpenClaw | Single binary / Rust | Simple loop / Python |
| Language | TypeScript | Python | TypeScript | Rust | TS + Python | Rust | Python |
| Codebase size | 500K lines | ~8,900 lines core | ~700 lines | Rust, auditable | Wrapper | 3.4MB binary | 4,000 lines |
| Self-improving / learns | No | Yes | No | No | No | No | No |
| Persistent memory | Context file | Yes, native | Per-container | Encrypted DB | Via OpenClaw | Partial | Basic |
| Messaging channels | |||||||
| Yes | Yes | Yes | Via plugins | Via OpenClaw | Partial | Via config | |
| Telegram | Yes | Yes | Yes | Yes | Via OpenClaw | Yes | Yes |
| Slack / Discord | Yes | Yes | Yes | Yes | Via OpenClaw | Partial | Yes |
| Signal | Yes | Yes | Partial | No | Via OpenClaw | No | No |
| iMessage | Yes | No | No | No | Via OpenClaw | No | No |
| Security | |||||||
| Container / sandbox isolation | No | 6 backends | Yes, per-agent | WASM + TEE | Kernel sandbox | Process only | No |
| Prompt injection scanning | Patched CVEs | Yes v0.7+ | Via isolation | iron-verify | Yes | Basic | No |
| Known critical CVEs | 3 critical | None | None | None | None | None | None |
| Suitable for regulated industries | No | With hardening | With hardening | Yes | Yes | Partial | No |
| Ecosystem & cost | |||||||
| Skills / plugin marketplace | ClawHub 5,700+ | agentskills.io | Fork-based | No | Via OpenClaw | No | No |
| MCP support | Yes | Yes | Via extension | Via plugins | Via OpenClaw | No | No |
| Local model support (Ollama) | Yes | Yes | Via Claude SDK | Yes | Nemotron bundled | Yes | Yes |
| Managed hosting available | Yes ($59/mo) | Partners (~$14/mo) | Self-host only | Enterprise | N/A | Self-host only | Self-host only |
| Runs on low-power hardware | 1GB+ RAM | VPS min. | Yes | Moderate | Host only | Yes | Raspberry Pi |
The lineage in one sentence: OpenClaw proved the category existed. NanoClaw and IronClaw were built because OpenClaw's security was dangerous. Hermes was built because OpenClaw didn't learn. ZeroClaw and Nanobot were built because OpenClaw was too heavy. NemoClaw was built because NVIDIA saw the enterprise opportunity. HiClaw was built because nobody had coordinated multiple agents yet.
Section II
Coding Agents, Managed Platforms
& Developer Frameworks
The broader agent landscape: terminal-native coding tools, hosted infrastructure for production deployment, big tech model-native agent systems, and the developer frameworks used to build custom architectures.
Coding agents — terminal-native developer tools
Mario Zechner
Minimal terminal coding harness. Only 4 core tools: read, write, edit, bash. Maximally extensible via TypeScript extensions. The engine that powers OpenClaw. Model-agnostic across 15+ providers. Mid-session model switching.
Anthropic
Batteries-included CLI. Deep codebase context, git workflows, IDE integration (VS Code, JetBrains, Xcode). Included in Pro/Max plans. SOC2 Type II. Foundation for Managed Agents. Best-in-class for Claude models.
xAI (now SpaceX)
Local-first CLI coding agent. Up to 8 parallel agents on a single project. Air-gap compatible, no code sent to xAI servers. Native X/real-time data advantage. Waitlist as of Q1 2026.
Block
Open-source local agent from Block. CLI and desktop, MCP extensibility, real engineering task automation. Model-agnostic. Privacy-first Claude Code alternative.
Community
Git-native CLI coding agent. Excellent multi-file editing, strong git integration. Simpler than Pi. Best for developers who want AI-assisted commits without setup overhead or extension work.
Terminal coding agent with A2A protocol and Vertex AI backend. Native Google Workspace and Search grounding. Strong for GCP workflows. Part of Google ADK ecosystem.
Managed agent platforms — hosted production infrastructure
Anthropic
Hosted harness with durable session logs, sandboxed cloud containers, crash recovery via wake(sessionId), credential vaults where secrets never reach the sandbox, and MCP tool integration. Brain decoupled from hands. The architecture described in Anthropic's engineering blog.
OpenAI
Managed runtime with built-in web search, code interpreter, file search, agent handoffs, and guardrails. Tightly OpenAI-coupled. Lowest setup friction for GPT-based deployments. Production-ready.
Enterprise agent platform on GCP. A2A protocol, Gemini models, native Google Workspace connectors. Strong for organisations already in GCP. Managed compliance, observability, and data residency.
Microsoft
No-code/low-code agent builder on M365. Deep Teams, Outlook, Dynamics integration. Now powered by Claude (Cowork partnership). Best for M365 organisations. Enterprise governance built-in.
Lindy AI
No-code personal agent platform. Connects Gmail, Calendar, Slack, CRMs. Automations from natural language. Consumer-friendly entry for non-developers wanting agentic workflows without self-hosting.
Perplexity Computer
perplexity.ai/products/computer ↗Perplexity AI
Multi-model cloud agent orchestrating 19 AI models simultaneously — routing each subtask to the optimal model. Tasks run in isolated Firecracker microVMs and can persist for hours or months. Also ships Personal Computer, a local variant running on Mac Mini for file and app access. Model-agnostic by design: no single provider dominates.
Big tech — model-native and internal agent systems
xAI (now SpaceX)
4-agent council baked into inference at runtime: Grok (Captain), Harper (research), Benjamin (math/code), Lucas (strategy). Not a framework you orchestrate — runs natively on every complex query. 16 agents on high reasoning mode.
Meta
Internal agent framework powering Meta's Ranking Engineer Agent. Multi-week autonomous ML experimentation via hibernate-and-wake mechanism. Not publicly available — internal Meta infrastructure only.
Microsoft Research
Conversation-based multi-agent framework. Agents communicate via structured chat. Strong human-in-the-loop. Azure integration. Open source and enterprise-deployable. v0.4 rewrite introduced graph-based workflows.
Developer frameworks — build your own agent architecture
LangChain Inc.
Largest ecosystem. LangGraph adds graph-based stateful workflows with explicit control flow and state management. Most integrations. LangSmith for observability. High flexibility, high complexity. Best for custom multi-step pipelines.
CrewAI Inc.
Role-based multi-agent orchestration. Define a crew of specialist agents with task delegation. Sequential or hierarchical coordination. Intuitive mental model for team-oriented workflows. Fast prototyping.
Agent Development Kit. Python-first, multi-agent, A2A protocol. Connects to Vertex AI and Google Workspace data stores. Designed for developers building on Google infrastructure.
LlamaIndex
RAG-first agent framework. Best for agents that need to reason over large document corpora. Deep indexing and retrieval primitives. Combines well with CrewAI for document-heavy multi-agent systems.
Community
TypeScript-first agent framework for web developers. Built-in Studio with traces and token usage. Durable execution, state management, and long-running workflow support. Growing fast in 2026.
Platform comparison
| Pi | Claude Code | Grok Build | Claude Managed | OpenAI Agents | Copilot Studio | Grok Multi-Agent | LangGraph | CrewAI | ||
|---|---|---|---|---|---|---|---|---|---|---|
| What it is | ||||||||||
| Category | Coding harness | Coding agent | Coding agent | Managed infra | Managed infra | Enterprise platform | Model-native | Dev framework | Dev framework | |
| Primary audience | Power devs | Devs / builders | Local-first devs | Production teams | GPT builders | Enterprise / M365 | Research / power users | Custom pipelines | Role-based crews | |
| Deployment | ||||||||||
| Self-hosted / local | Yes | Yes | Yes | No | No | No | N/A | Yes | Yes | No |
| Runs 24/7 background | No | No | No | Yes | No | Yes | N/A | No | No | Yes |
| Model agnostic | Yes, 15+ | Claude only | Grok only | Claude only | OpenAI primary | Claude + GPT | Grok only | Yes | Yes | Yes, 19 models |
| Data sovereignty | Yes | Local + Anthropic | Yes | No | No | No | No | Yes | Yes | No |
| Capabilities | ||||||||||
| Multi-agent orchestration | Via extension | Yes | 8–16 parallel | Preview | Yes | Yes | Yes, native | Yes | Yes | Yes, native |
| MCP support | Via extension | Yes | No | Yes | No | No | No | Via plugins | Via plugins | No |
| Real-time web / data | Via extension | Web search | Yes + X firehose | Yes | Yes | Yes | Yes, native X | Via tools | Via tools | Yes, Sonar |
| Code execution (sandboxed) | Yes | Yes | Local | Cloud sandbox | Yes | Limited | N/A | Local | Local | |
| Production architecture | ||||||||||
| Durable session / crash recovery | Tree branching | Partial | No | Yes, durable log | Partial | Yes | N/A | Setup needed | No | Yes |
| Credential vault (secrets from sandbox) | No | No | Local | Yes | Yes | Yes | N/A | No | No | |
| Observability / tracing | Session tree | Partial | Auditable local | Yes | Yes | Yes | N/A | LangSmith | Enterprise | Yes |
| Cost & access | ||||||||||
| Free beyond API costs | Yes | $20/mo | Waitlist | +$0.08/hr | Token-based | M365 licence | Via subscription | Yes | Yes | $200/mo |
| Non-developer accessible | No | No | No | Console UI | No | Yes | Yes | No | No | Yes |
| Security / compliance | MIT, unaudited | SOC2 Type II | Local, unaudited | SOC2 Type II | SOC2 Type II | Enterprise | xAI / SpaceX | Open source | Open source | Sandbox + human gates |
The four-layer model: The model layer (Claude, Grok, Gemini) provides intelligence. The harness layer (Pi, Claude Code, Grok Build) provides the execution loop. The orchestration layer (OpenClaw, Hermes, LangGraph, CrewAI) routes tasks and manages state. The managed infrastructure layer (Claude Managed Agents, OpenAI Agents SDK, Copilot Studio) handles production deployment, security, and crash recovery. Most people conflate these layers. The real competitive advantage lies in the orchestration layer — not the model.