Enterprise-Class Architecture

Built for Fortune 500 reliability. Engineered with modern frameworks, defense-in-depth security, and production-proven design patterns that scripts and scheduled scripts simply cannot match.

Modern Application Framework

InTouch AI is built on the Micronaut framework — the same technology chosen by leading cloud-native platforms for its performance, security, and scalability characteristics.

Micronaut 4.x

Modern JVM framework with compile-time dependency injection. No reflection, no runtime proxies, no classpath scanning. Fast startup, low memory footprint, predictable behavior.

Compile-Time DI

All dependency injection resolved at compile time — not runtime. This eliminates entire classes of runtime errors, reduces attack surface, and enables ahead-of-time (AOT) compilation.

Runs Anywhere

Single JAR, standard JVM 17+. Sub-second startup, minimal memory. Runs on everything from a Raspberry Pi to enterprise data center servers — Linux, Windows, macOS, ARM.

OpenAPI 3.0

Complete API specification auto-generated from code annotations. Integrated Swagger UI for live API testing. Auto-generated client SDKs for any language.

Reactive & Non-Blocking

Built on Jetty — the battle-tested HTTP engine behind Eclipse, Jenkins, and Google App Engine. Reliable, proven, handles thousands of concurrent API requests and tool executions.

Platform Independence

Pure Java application runs anywhere the JVM runs — Linux, Windows, macOS, ARM, x86. No native dependencies, no OS-specific installation procedures.

Defense-in-Depth Security

Every layer of InTouch is designed with security as a first-class concern — not an afterthought bolted on.

Credential Management

  • AES-256 encryption for all stored credentials
  • Centralized credential vault — no passwords in scripts, config files, or environment variables
  • CyberArk integration — enterprise privileged access management (Department & Enterprise editions)
  • Connection-level security — credentials never exposed to tool authors
  • Encrypted at rest and in transit — credentials never stored in plaintext anywhere

Access Control

  • Role-Based Access Control (RBAC) — granular permissions by role, project, and object
  • LDAP / Active Directory — enterprise identity provider integration
  • Session-based authentication — secure token management with automatic expiration
  • Object-level security — control access to individual jobs, connections, and projects
  • Full audit trail — every action logged with user, timestamp, and detail

REST API Architecture

The InTouch REST API is not a wrapper around an internal interface — it is the interface. The same API that powers the InTouch web UI is available to external integrations, ensuring complete feature parity.

Endpoints
413 across 26 domains
Specification
OpenAPI 3.0 compliant
Documentation
Integrated Swagger UI
Authentication
Session-based with RBAC
Formats
JSON request/response
MCP Server
367 tools for Claude Code

API-First Design

Every capability in InTouch is accessible through the REST API. This enables:

  • Self-service automation portals
  • CI/CD pipeline integration
  • Cloud function triggers (AWS Lambda, Azure Functions)
  • Custom monitoring dashboards
  • Mobile application integration
  • Natural language automation via MCP

View API Technical Reference

Intelligent Orchestration Engine

Hierarchical Organization

Project > Jobset > Job > Tools — a clean organizational model that scales from single-user automation to enterprise-wide deployment with thousands of jobs.

Multi-Type Tool Sequences

Each job can contain multiple tools of different types — SQL extraction, DataFrame transformation, Claude analysis, FTP distribution — executed in sequence with automatic error handling.

Flexible Triggering

7 native schedule types (day, week, weekday, weekend, month/specific, month/relative, custom), file-based triggers (detect new files), event-driven triggers, ad-hoc OneShot runs (REST API or UI with parameter overrides), and AI triggers that evaluate conditions in natural language.

Mutual Exclusion

Prevent concurrent execution on shared resources. If a job is already running against a database or cube, new executions queue automatically — no race conditions, no data corruption.

Chase-Its-Tail Mode

Continuous execution mode that immediately restarts upon completion. Process incoming work as fast as possible with zero idle time between cycles.

Multi-Channel Notifications & Alerts

Eight outbound delivery channels: Email (SMTP), Slack, Discord, Telegram, SMS (Twilio/Telnyx/Plivo), WhatsApp, Teams, and LINE. Per-job notifications plus standalone alerts linked to triggers. Same credential vault, same audit log, same RBAC across every channel.

The Cost of DIY Automation

Many organizations start with "simple" scripts and scheduled scripts. Here's what that simplicity actually costs:

The DIY Scripts Approach

The typical DIY automation stack — shell scripts, Python scripts, scheduled scripts, and Jenkins pipelines — creates a growing liability that becomes increasingly dangerous and expensive over time.

ConcernDIY ScriptsInTouch AI
Credentials Plaintext passwords in scripts, config files, environment variables, or .env files. Committed to git repos. Visible to anyone with server access. One breach exposes everything. AES-256 encrypted vault. CyberArk integration. Credentials never exposed to tool authors. Never in plaintext anywhere on disk.
Access Control Linux file permissions at best. Anyone with server SSH access can read, modify, or delete any script. No per-job or per-resource permissions. Full RBAC with roles, object-level permissions, LDAP/Active Directory integration. Granular control over who can see, edit, or execute each job.
Error Handling Silent failures. Scripts exit with errors that nobody sees. Failures discovered hours or days later — or not at all. Immediate notification via email, SMS, or Slack. Automatic retry. Complete execution logs with timestamps. Live monitoring dashboard (Enterprise).
Audit Trail None. No record of who ran what, when, or what changed. Compliance auditors have nothing to review. Complete audit trail of every execution, every configuration change, every user action. Timestamped and attributable.
Knowledge Transfer Tribal knowledge. Undocumented scripts written by employees who have left. No one knows what they do, but everyone is afraid to touch them. Self-documenting job definitions with descriptions. Visual job hierarchy. Any administrator can understand and manage any workflow.
Dependency Management Python version conflicts. pip vs conda. Node version mismatches. "Works on my machine." Production breaks after OS updates. Self-contained Java application. Automatic Maven dependency resolution for DataFrame tool runs. No external runtime dependencies.
Scheduling Schedule files scattered across servers. No dependency chains. No visual schedule overview. No way to pause and resume. Visual scheduler with dependency chains, file triggers, event triggers, AI triggers, and OneShot ad-hoc runs with parameter overrides.
Scalability Each new automation is another script, another scheduled script, another thing to maintain. Linear growth in operational burden. Add new jobs through the UI or API. Same management overhead whether you have 10 jobs or 10,000.
AI Integration Write custom code for every AI API call. Manage API keys in environment variables. No error handling, no retry, no logging. 7 native AI tools. Scheduled, triggered, monitored, and error-handled. API keys in encrypted vault. AI triggers and autonomous agents.

Security Risks of DIY Automation

The Plaintext Password Problem

A 2023 GitGuardian report found 10 million secrets leaked in public GitHub repositories in a single year. Enterprise private repos are no better — database passwords in shell scripts, API keys in Python files, connection strings in .env files.

One compromised server, one careless git push, one backup tape — and every credential in every script is exposed. With InTouch, credentials exist only in the AES-256 encrypted vault and are never accessible in plaintext, even to server administrators.

The Silent Failure Problem

The DIY error-handling model is the same one shell scripts have always had — fail silently, write to a log nobody reads, and hope. In practice, this means:

  • Root's mailbox overflows and emails are silently discarded
  • Scripts that exit with errors produce no notification
  • Partial failures (3 of 5 steps succeed) go completely undetected
  • Nobody knows a job failed until a downstream consumer complains
  • By then, data integrity may be irreparably compromised

InTouch provides immediate notification on any failure, with complete execution logs showing exactly what happened, when, and why.

InTouch vs. General-Purpose Tools

Jenkins, Airflow, and other pipeline tools are designed for software developers building CI/CD pipelines — not for business users automating enterprise processes.

RequirementJenkins / AirflowInTouch AI
Setup Complexity Requires DevOps expertise. Python DAGs (Airflow) or Groovy pipelines (Jenkins). Plugin ecosystem with version conflicts. Single Java application. Configure through web UI or REST API. No coding required for standard automation.
Enterprise Systems No native Essbase, TM1, or JDE tools. Must write custom integrations from scratch. Purpose-built tools for Essbase (9 operations), TM1 (fanatical session management), and JDE with extensive production hardening.
Database Operations Generic database hooks. No proprietary batching. No SQL Stream. No visual query building. 5 SQL operation types with 1000X faster imports. Direct database-to-database streaming. 14+ database support.
AI Integration Third-party plugins or custom Python code. No native AI tools. No AI triggers. No autonomous agents. 17 native AI task types across 9 providers. AI triggers. Autonomous Claude Agent. 367-tool MCP server. Built-in AI assistant.
Target Users Software engineers and DevOps teams Business users, DBAs, analysts, IT operations — plus developers via API
Operational Model Requires dedicated team to maintain infrastructure, update plugins, manage dependencies. Self-contained application. Updates are a single file replacement. Minimal operational overhead.

Technical Specifications

Application Framework
Micronaut 4.7.6 with compile-time DI (KSP)
Runtime
Java 17+ (any JVM); Kotlin 1.9.25
HTTP Server
Jetty (battle-tested, reliable)
API Specification
OpenAPI 3.0 with Swagger UI
Encryption
AES-256 for credentials at rest
Authentication
Session-based, LDAP, Active Directory
Authorization
RBAC with object-level permissions
Credential Vault
Built-in + CyberArk integration
Database Support
MySQL, MariaDB, Oracle, PostgreSQL, SQL Server, DB2, Firebird, Cloud Spanner, Informix, Derby, H2, SQLite, Sybase, Cassandra
Supported Platforms
Linux, Windows, macOS, ARM, x86 — anywhere Java runs
Minimum Hardware
Raspberry Pi (512MB RAM)
Tools
42+ built-in tools: 18 JAR plugins, 13 core tools, 12+ YAML tools. Hot-reload without server restart.
AI Providers
Anthropic Claude, OpenAI, Mistral, Groq, DeepSeek, xAI, Google Gemini, Ollama (8 native — all editions). Hugging Face as a 9th, job-only.
Skills
Native InTouch skills (SKILL.md) + 5,000+ OpenClaw skills from upstream ClawHub (auto-install on first use, converted to deterministic YAML)
Outbound Notifications
8 outbound delivery channels: Email (SMTP), Slack, Discord, Telegram, SMS (Twilio/Telnyx/Plivo), WhatsApp, Teams, LINE.
REST API
413 endpoints, 26 functional domains
MCP Server
367 tools for Claude Code integration
Editions
Personal (free), Team, Department, Enterprise

Enterprise Architecture Without the Enterprise Price Tag

Start free with the Personal edition. Scale to Enterprise when you need RBAC, LDAP, and live monitoring.

Contact Blue Isle Software Compare Editions