InTouch AI vs OpenClaw — Comparison Sheet
From Individual to Enterprise — One Platform, Every Scale
If you're using OpenClaw today: a great single-user AI agent on a laptop — LLM tokens spent on every run, no audit log, no RBAC, no enterprise scaling. The skills ecosystem (5,000+ skills) is the draw, but the runtime is single-user and per-run AI cost adds up fast. You'd love to keep the skills catalogue without the single-machine ceiling and the per-execution token bill.
What makes InTouch AI unique: InTouch is self-hostable (free Personal edition), single-tenant by design, runs behind NAT via Tailscale, has a real governance layer (RBAC + encryption + audit), and agentic AI built in — not bolted on. It ships with 42+ tools, eight outbound notification channels (Email, Slack, Discord, Telegram, SMS, WhatsApp, Teams, LINE), Job Files, and runs both OpenClaw skills (5,000+ from upstream ClawHub, always converted to deterministic YAML before execution) and InTouch-native MD skills.
Governance: OpenClaw Standalone vs InTouch
| Governance Layer | OpenClaw (standalone) | InTouch |
|---|
| Access control | None — user runs everything | RBAC — roles, projects, publisher permissions |
| Credentials | Shell env vars (export API_KEY=…) | Encrypted vault, never exposed to AI context |
| Audit trail | None | Full job log with timestamps, inputs, outputs |
| Scheduling | Manual invocation only | 7 native schedule types, file/folder/AI triggers, ad-hoc OneShot runs |
| Alerting | None | Subscriber notifications on success/failure across 8 channels |
| Concurrency | None | Collision detection, exclusive job locks |
| Ownership | Single user on a local machine | Owner + group + per-role rights |
| Tenancy model | Single user, single machine | Single-tenant by design (one server per organization). Department/Enterprise add project isolation within the tenant. |
| Execution cost | LLM tokens on every run | Zero — converted skills run deterministically |
Why it matters: OpenClaw skills are never run raw in InTouch — they're always converted to a native YAML tool or YAML job first. Subsequent runs are zero-LLM-cost, deterministic, and audit-logged. ClawHub is a discovery source; InTouch is the governed runtime.
Self-automation closes the loop: InTouch's
intouch_publisher credential lets a runtime env script call the InTouch REST API as a real publisher — with that publisher's RBAC and audit trail — without ever holding the password. So a discovered ClawHub skill, once converted, can also be migrated, deployed, monitored, or rewritten by another script running under the same governance.
See self-automation.html.
Overview
| InTouch AI | OpenClaw |
|---|
| Type | Full-spectrum job automation — individuals to enterprise | Open-source personal AI assistant/agent |
| Architecture | Server-based (JVM/Kotlin/Micronaut), multi-user capable | Node.js agent runtime, single-user per instance |
| Primary Focus | Job scheduling, workflow automation, ETL, AI-native with 42+ tools | AI-powered personal automation, chat-driven tasks |
| Installation | Seconds — single JAR, all editions | Minutes (Node.js, npm dependencies) |
| Licensing | Free Personal edition; licensed Team, Dept, Enterprise | Open-source (Apache 2.0), pay for AI API usage |
Editions at a Glance
| Feature Area | InTouch Personal (Free) | InTouch Team | InTouch Dept | InTouch Enterprise | OpenClaw |
|---|
| Price | Free — forever | Licensed | Licensed | Licensed | Free + API costs ($5–100+/mo) |
| All Tools | Yes — all 42+ tools | Yes | Yes | Yes | AI-driven only |
| Users | 1 (auto-login) | License-limited | License-limited | License-limited | 1 per instance |
| Access Control | None needed | All admins | Full RBAC + LDAP | Full RBAC + LDAP | None |
Tool Execution
| Capability | InTouch AI (All Editions) | OpenClaw |
|---|
| Tools | 42+ built-in: SQL (9 shipped + 4 drop-in), FTP/SFTP, SSH, HTTP, AWS (full CLI v2), Google Workspace, Docker, AI (Anthropic x5, OpenAI, Gemini, Ollama), Excel, PDF, Git, LDAP, MongoDB, Cassandra, DataFrame, Essbase, TM1, JDE, Email, Message, File Mgmt, Runtime Env, OpenClaw + 12+ YAML tools | AI-driven tasks via skills (100+ AgentSkills) |
| Plugin System | 18 IToolConnector JAR plugins with JSON Forms schema for dynamic UI | AgentSkills |
| Job Composition | Multi-step jobs with property passing. Job Files with dependency DAGs. | Single-action scheduled scripts; sub-agent chaining |
| Job Files | Jobs-as-code with dependency DAGs, property interpolation, failure handling | Not available |
| Variable Resolution | Built-in resolver with job/system/date variables | Template variables in prompts |
Skills: OpenClaw vs InTouch MD
| Capability | InTouch AI | OpenClaw |
|---|
| OpenClaw Skills | Runs 5,000+ ClawHub skills without OpenClaw runtime. Auto-install on first use. | Native skill system |
| MD Skills | InTouch-native markdown skills that orchestrate tools via AI. Separate from OpenClaw. | Not applicable |
| Invocation | From the AI assistant chat (web UI, PWA, Android), REST API, scheduled, or as a job tool | Via chat or schedule |
| Auto-Install | Skill not found? Searches ClawHub (5,000+ upstream OpenClaw), installs, runs automatically. | Manual installation |
| Skill as Job Tool | OpenClaw tool — run skills in automated jobs with schedules/triggers | Via scheduled scheduled scripts |
Scheduling & Triggers
| Capability | InTouch AI (All Editions) | OpenClaw |
|---|
| Schedule Types | 7 native types: Day, Week, Weekday, Weekend, Month/Specific, Month/Relative, Custom | 5-field schedule expressions |
| File Triggers | Built-in file monitoring with settle time | Not built-in |
| AI Triggers | LLM evaluates conditions to trigger jobs | Not built-in |
| Alert System | Dedicated alert entities with multi-channel notifications | Not available |
| Blackout Dates | Built-in holiday/blackout support | Not available |
Outbound Notification Channels
| Capability | InTouch AI | OpenClaw |
|---|
| Channels | 8 outbound: Email (SMTP), Slack, Discord, Telegram, SMS (Twilio/Telnyx/Plivo), WhatsApp (Meta Cloud API), Microsoft Teams, LINE | Chat platform delivery only |
| Notifications | Per-subscriber routing across all 8 channels on job events | Chat platform delivery |
AI Capabilities
| Capability | InTouch AI | OpenClaw |
|---|
| Agentic Assistant | 76 tool_use functions for natural language automation | Entire platform is AI assistant |
| AI Providers | Anthropic Claude (5 types), OpenAI, Gemini, Ollama | Claude, GPT, Gemini, DeepSeek, Grok, Ollama |
| AI in Workflows | AI tools chain with SQL, FTP, SSH, HTTP — structured workflows | AI orchestrates via natural language |
| Batch AI | Anthropic Batch API support | Not built-in |
| AI Safety | Mandatory safety preamble on all AI calls | No built-in safety controls |
| MCP Server | Built-in for Claude Code and other AI tools | MCP client support |
Summary: When to Choose What
| Choose InTouch AI When You Need | Choose OpenClaw When You Need |
|---|
| A platform that grows with you — free for individuals, scales to enterprise | A personal AI assistant for ad-hoc automation |
| 42+ built-in tools (SQL, FTP, SSH, AWS, enterprise, data tools) | Natural language driven tasks on your local machine |
| Agentic AI assistant with 76 tool_use functions | Chat platform integration (WhatsApp, Discord, Telegram) |
| 8 outbound delivery channels (5 on every edition; 8 on Department/Enterprise) | Quick AI-powered scripts and browser automation |
| Both OpenClaw skills (5,000+) AND InTouch MD skills | Native OpenClaw skill ecosystem |
| Job Files for jobs-as-code with dependency DAGs | Lightweight scheduling with AI model flexibility |
| Enterprise-ready RBAC, LDAP, and audit trails | Open-source with community extensibility |
| Production-grade monitoring, alerts, multi-channel notifications | Personal productivity and chat-based workflows |