Hybrid Architecture

Heptora’s Hybrid Architecture is an innovative model that combines the best of both worlds: the power and scalability of the cloud with the security and privacy of local processing. This architecture ensures that your sensitive data never leaves your infrastructure while you take advantage of advanced AI and cloud orchestration capabilities.

Overview

In a world where data privacy is increasingly critical and regulations like GDPR are strict, Heptora’s Hybrid Architecture offers a unique solution: powerful automation without compromising your data security.

Why a Hybrid Architecture?

🔒 Guaranteed Privacy: Sensitive data never leaves your infrastructure
☁️ Cloud Power: Leverage AI and advanced orchestration capabilities
⚖️ Regulatory Compliance: Facilitates compliance with GDPR, CCPA and other regulations
🚀 Best of Both Worlds: Combines local security with cloud innovation
💪 Total Control: You decide what data stays local and what the cloud processes
🔄 Risk-Free Updates: Platform improvements without compromising security

The Two Components

Cloud Component: The Web Platform

Heptora’s cloud component is a modern and powerful web application that acts as the command and control center for all your automations.

Cloud Component Responsibilities

1. Process Orchestration

The cloud component coordinates the execution of your automations:

Centralized scheduling: Defines when and how processes execute
Multi-robot coordination: Manages multiple robots working together
Intelligent sequencing: Executes processes in optimal order
Dependency management: Ensures processes execute in correct order
Load balancing: Distributes work among available robots

Orchestration example:

Process: Monthly Report Generation

Cloud Orchestrator:
  1. Schedule start on 1st of each month at 8:00 AM
  2. Verify Robot-Finance is available
  3. Queue task "Extract ERP data"
  4. When complete, queue "Process and analyze"
  5. When complete, queue "Generate PDF report"
  6. When complete, queue "Send via email"
  7. Notify user of final result

2. Task Queuing System

Efficient workload management:

Priority queues: Urgent tasks are processed first
Capacity management: Prevents robot overload
Intelligent retries: Retries failed tasks with exponential logic
Persistent state: Never loses tasks even with technical issues
Complete visibility: See all pending, in-process and completed tasks

Queuing flow:

New Task
    ↓
Validation → [Reject if invalid]
    ↓
Assign Priority
    ↓
Queue → [High Priority Queue / Normal Queue / Low Priority Queue]
    ↓
Available Robot → Takes task from queue
    ↓
Executes Locally
    ↓
Reports Result → Cloud updates status

3. Artificial Intelligence Assistants

The power of modern AI applied to automation:

Construction Assistant: Helps design processes visually
Code Assistant: Generates custom Python from natural language
Diagnostic Assistant: Analyzes errors and suggests solutions
Onboarding Assistant: Guides new users step by step

These assistants require very large AI models and significant processing power, so they live in the cloud. However, they never process your business data, only metadata and descriptions.

What AI Assistants see:

✓ "I need to extract emails from text"
✓ "Process failed with error: ConnectionTimeout"
✓ Process structure (blocks and connections)
✓ Type of data each block handles

✗ Actual content of your invoices
✗ Your customer names
✗ Personal or sensitive data
✗ Access credentials

4. User Interface and Management

Complete web portal to manage your automation:

Interactive dashboard: Visualize status of all your processes
Visual process editor: Design automations with drag-and-drop
User and permission management: Role-based access control
Real-time monitoring: See what your robots are doing right now
Reports and analytics: Measure your automation performance
Template management: Catalog of pre-built processes ready to use

5. Metadata Storage

Information about processes, but not the processed data:

Process definitions: How each process is built
Execution history: When it ran, how long it took, result
Structured logs: Activity records (without sensitive data)
Configurations: General settings for each process
Versions: Change history in processes

What’s NOT in the Cloud

It’s equally important to understand what is NOT stored or processed in the cloud:

❌ Processed business data
❌ File content
❌ Credentials and passwords
❌ Personally identifiable information
❌ Sensitive financial information
❌ Business documents
❌ Secrets and access tokens

Local Component: The Execution Robot

The local robot is a software agent that runs in your infrastructure and performs the actual automation work.

Local Component Responsibilities

1. Process Execution

The robot executes automations where it really matters: in your environment:

Local processing: All execution occurs on your machine
Access to local systems: Can interact with applications on your network
File manipulation: Reads and writes files in your folders/servers
Application control: Automates software on your desktop
Database connection: Accesses your databases directly

Local execution example:

Local Robot (on your server):
  1. Receives instruction from cloud: "Execute Invoicing process"
  2. Reads XML invoice files from \\local-server\invoices\
  3. Accesses local database with locally stored credentials
  4. Processes each invoice (validation, transformation, etc.)
  5. Writes results to local folder \\local-server\results\
  6. Reports to cloud: "Processed 47 invoices successfully"
     (without sending invoice content)

2. Sensitive Data Storage

The robot keeps all confidential information securely local:

Processed files: Remain in your file systems
Databases: Direct access without passing through internet
Documents: PDFs, Excel, Word remain local
Intermediate results: Temporary data during processing
Caches and detailed logs: Complete technical information local

3. Global Variable Management

Variables shared between processes, stored locally:

# Global variables (stored locally)
SHARED_PATH = "\\server\shared\data\"
NOTIFICATIONS_EMAIL = "operations@company.com"
DEFAULT_TIMEOUT = 30
ENVIRONMENT = "production"

These variables can be used by multiple processes but never leave your infrastructure.

4. Secrets Management

The robot includes an enterprise-grade secrets management system:

Encrypted storage: All secrets are encrypted at rest
AES-256 encryption: Military-grade data protection standard
Controlled access: Only the robot can decrypt
No transmission: Secrets are never sent to the cloud
Key rotation: Ability to change passwords without breaking processes

Types of secrets stored:

System credentials:
  - ERP username/password
  - CRM username/password
  - Digital certificates
  - API tokens

Database connections:
  - SQL Server connection strings
  - MongoDB credentials
  - Local database access

Service access:
  - API keys
  - OAuth tokens
  - SSH keys
  - SSL/TLS certificates

5. Secure Communication with Cloud

The robot maintains a secure connection with the cloud component:

TLS/SSL: All communication is encrypted
Mutual authentication: Robot and cloud verify identities
Regular heartbeat: Robot periodically reports it’s alive
Metadata only: Communication contains only control information

Communication protocol:

Robot → Cloud: [Encrypted] "Process123 started"
Cloud → Robot: [Encrypted] "OK, waiting for result"

[Robot executes process locally with local data]

Robot → Cloud: [Encrypted] "Process123 completed.
                           Duration: 5min.
                           Records processed: 142.
                           Status: Success"
Cloud → Robot: [Encrypted] "Recorded. Next task: Process456"

Note: Content of the 142 records is not transmitted, only aggregated metadata.

Robot Deployment

The robot can be deployed in multiple ways:

1. Workstation Installation

Robot on individual PC or laptop
Ideal for desktop processes
Easy configuration for individual users

2. Server Installation

Robot on Windows/Linux server
Multiple concurrent processes
Greater capacity and reliability
Ideal for enterprise production

3. Virtual Machine Installation

Robot on dedicated VM
Additional isolation and security
Easy backup and replication
Scaling through more VMs

4. Containerization (Docker)

Robot in Docker container
Maximum portability
Integration with orchestrators (Kubernetes)
Ideal for modern architectures

Complete Workflow

Let’s see how both components interact in a real example:

Example: Automatic Invoice Processing

Phase 1: Design (in Cloud)

User in web browser:
1. Access app.heptora.com
2. Use visual editor to create process
3. Use AI Assistant to generate custom block
4. Configure schedule: Daily at 10:00 PM
5. Save process

Phase 2: Configuration (Local)

User in local robot interface:
1. Configure secret: SAP credentials
2. Define global variables:
   - INVOICES_FOLDER = "C:\Invoices\Incoming"
   - PROCESSED_FOLDER = "C:\Invoices\Processed"
3. Map connection: Local database

Phase 3: Execution (Hybrid)

10:00 PM - Cloud Orchestrator:
  "Time to execute Invoices process"
  → Queue task for Robot-Server01

Robot-Server01:
  ← Receives instruction: "Execute Invoices process"

  → Access INVOICES_FOLDER (local)
  → Read 23 PDF files (local)
  → For each invoice:
       Use local OCR to extract data
       Retrieve SAP credentials from local secrets
       Connect to SAP (local/VPN)
       Register invoice in SAP
       Move PDF to PROCESSED_FOLDER (local)
  → Generate detailed log (stored locally)

  → Report to cloud:
       "Process completed.
        23 invoices processed.
        Duration: 8 minutes.
        Status: Success"

Cloud:
  ← Receive report
  → Update dashboard
  → Send notification to user (via cloud)
  → Record aggregated metrics

Phase 4: Monitoring (in Cloud)

User in browser:
1. See updated dashboard: "23 invoices processed"
2. Review execution timeline
3. See metrics: Average time, success rate
4. Does NOT see invoice content (it's only local)

What Traveled to the Cloud?

In this entire process, the information that went to the cloud was:

Metadata sent to cloud:
✓ "Process 'Invoices' started at 10:00:03 PM"
✓ "23 files detected for processing"
✓ "File 1/23 processed correctly"
✓ "File 2/23 processed correctly"
...
✓ "Process completed at 10:08:15 PM"
✓ "Total duration: 8 minutes 12 seconds"
✓ "100% success"

Data NOT sent to cloud:
✗ PDF content
✗ Extracted data (amounts, dates, suppliers)
✗ SAP credentials
✗ Detailed logs with business information
✗ Specific file paths

Hybrid Architecture Advantages

1. Privacy and Security

Simplified Regulatory Compliance

Hybrid architecture dramatically facilitates regulatory compliance:

GDPR (General Data Protection Regulation):

✓ Personal data remains in your infrastructure under your control
✓ No international transfers of personal data
✓ You are the sole data controller for sensitive data
✓ Don’t need extensive DPAs (Data Processing Agreements) with Heptora for processed data
✓ Rights of access, rectification and deletion completely under your control

CCPA (California Consumer Privacy Act):

✓ Consumer data remains in your systems
✓ No sale of data to third parties
✓ Complete transparency on data usage
✓ Easy compliance with consumer requests

Sector-specific (Banking, Healthcare, etc.):

✓ Critical data never leaves your certified environment
✓ Complete local traceability
✓ Audit capability without depending on third parties

Protection Against Security Breaches

Even if there were a security issue in the cloud:

Hypothetical scenario: Breach in cloud component

Compromised information:
  - Process names
  - General configurations
  - Execution history
  - Users and permissions

Information NOT compromised (it's local):
  - Customer data
  - Invoice content
  - System credentials
  - Processed files
  - Detailed logs with business information

Environment Segregation

You can have multiple robots at different security levels:

Robot-DMZ (demilitarized zone):
  - Processes that interact with internet
  - Lower sensitivity level
  - Can be in private cloud

Robot-Core (internal network):
  - Processes with highly sensitive data
  - No direct internet access
  - In certified infrastructure

Robot-Production:
  - Critical business processes
  - Maximum availability
  - Intensive monitoring

Robot-Development:
  - Testing and experimentation
  - Non-sensitive test data
  - Greater flexibility

2. Optimized Performance

Heavy Data Processing

When handling large data volumes, local processing is much more efficient:

Comparison: Processing 1000 PDF invoices

Cloud-Only Architecture:
  1. Upload 1000 PDFs to cloud (total size: 500MB)
     → Time: 10-30 minutes (depending on bandwidth)
     → Cost: Data transfer
  2. Process in cloud
     → Time: 15 minutes
  3. Download results
     → Time: 5 minutes

  Total: ~50 minutes

Heptora Hybrid Architecture:
  1. Robot accesses PDFs locally
     → Time: instantaneous
  2. Process locally
     → Time: 15 minutes
  3. Report metadata to cloud
     → Time: < 1 second

  Total: ~15 minutes

Savings: 70% time

Reduced Latency

For interactive or high-frequency processes:

Local database access:
  - Latency: < 1ms
  - Throughput: Limited only by local hardware

vs

Database access via cloud:
  - Latency: 50-200ms (depending on location)
  - Throughput: Limited by internet bandwidth

Lower Internet Dependence

Scenario: 30-minute internet outage

Cloud-Only Architecture:
  ✗ All processes stop
  ✗ Total productivity loss
  ✗ Possible data loss in transit

Heptora Hybrid Architecture:
  ✓ Local processes continue executing
  ✓ Data keeps processing
  ✓ When connection restores, syncs metadata
  ✗ Cannot start new scheduled processes
  ✓ Minimal disruption

3. Flexibility and Control

Total Data Control

Decisions you control 100%:
  - Where processed data is stored
  - How long it's retained
  - Who has access (at OS level)
  - Backup and recovery policies
  - Additional encryption if needed
  - Storage geography

Deep Customization

By executing locally, you can:

Install any additional software needed
Use proprietary libraries and dependencies
Integrate with legacy systems without exposing them
Apply specific network configurations
Use specialized hardware (GPUs, etc.)

Multiple Environments

Easy management of development, testing and production:

Cloud (single):
  - Centralized management
  - Single interface

Local Robots (multiple):
  - Robot-Dev: Development and testing
    → Test data
    → Flexible configuration

  - Robot-QA: Quality control
    → Staging data
    → Pre-production configuration

  - Robot-Prod: Production
    → Real data
    → Production configuration
    → High availability

4. Optimized Cost

Transfer Cost Reduction

Typical monthly processing:
  - 10,000 invoices
  - Average 500KB per invoice
  - Total: 5GB of data

Cost in cloud-only architecture:
  - Upload: 5GB
  - Processing: 5GB
  - Download: 2GB results
  - Total transfer: 12GB
  - Cost: Variable by cloud provider

Cost in hybrid architecture:
  - Upload: ~0GB (metadata only)
  - Processing: Local
  - Download: ~0GB (metadata only)
  - Total transfer: < 100MB
  - Savings: > 99%

Use of Existing Infrastructure

You can reuse:
  ✓ Already available servers
  ✓ Workstations during off-hours
  ✓ Existing virtual machines
  ✓ Current backup infrastructure
  ✓ Current monitoring systems

You don't need:
  ✗ Contract additional cloud servers
  ✗ Pay for cloud storage of processed data
  ✗ Pay for massive bandwidth

Ideal Use Cases

1. Sensitive Information Processing

Financial Sector

Case: Bank processing loan applications

Data involved:
  - Personal information (ID, address, etc.)
  - Financial data (income, debts, accounts)
  - Documentation (tax returns, payslips)
  - Risk analysis (credit scoring)

Hybrid Architecture:
  Cloud:
    - Orchestrates application process
    - Manages approval workflow
    - Control panel for analysts

  Local:
    - Processes all documentation
    - Performs credit scoring
    - Accesses customer databases
    - Generates reports
    - ALL remains in bank's certified infrastructure

Healthcare Sector

Case: Hospital automating medical report processing

Data involved:
  - Medical records
  - Test results
  - Diagnoses
  - Patient personal data

Hybrid Architecture:
  Cloud:
    - Schedules nightly processes
    - Metrics dashboard (number of reports, times)
    - Process template management

  Local:
    - Processes medical reports
    - Extracts test data
    - Updates HIS (Hospital Information System)
    - ALL under local HIPAA/GDPR compliance

2. Integration with Legacy Systems

Case: Manufacturing company with legacy ERP without API

Infrastructure:
  - AS/400 ERP (no internet access)
  - Local database with no external exposure
  - Proprietary desktop applications

Hybrid Architecture:
  Cloud:
    - Modern interface for users
    - Synchronization scheduling
    - Reports and analytics

  Local:
    - Robot executes on internal network
    - Direct AS/400 access
    - Desktop application automation
    - Data extraction and transformation
    - NO need to expose ERP to internet

3. High Data Volume

Case: E-commerce processing 100,000 daily orders

Volume:
  - 100,000 CSV files with orders
  - 5GB daily data
  - Product image processing
  - Shipping label generation

Hybrid Architecture:
  Cloud:
    - Orchestrates distributed processing
    - Manages work queues
    - Real-time monitoring

  Local:
    - Multiple robots process in parallel
    - High-speed access to local storage
    - Image processing with local GPU
    - No data transfer limits
    - Results stored locally

4. Low Latency Requirements

Case: Automated trading

Requirements:
  - Millisecond response
  - Real-time market data access
  - Instantaneous decisions

Hybrid Architecture:
  Cloud:
    - Strategy management
    - Post-trading analysis
    - Performance reports

  Local:
    - Robot places orders locally
    - Direct broker access without internet latency
    - Local market data processing
    - Microsecond decisions

Deployment Considerations

Local Robot Requirements

Minimum Hardware

Light Use (1-5 small processes):
  - CPU: 2 cores
  - RAM: 4GB
  - Disk: 50GB SSD
  - Network: Stable internet connection

Medium Use (5-20 processes or OCR processing):
  - CPU: 4 cores
  - RAM: 8GB
  - Disk: 100GB SSD
  - Network: Stable internet connection

Intensive Use (20+ processes or massive processing):
  - CPU: 8+ cores
  - RAM: 16GB+
  - Disk: 250GB+ SSD (or fast network storage)
  - Network: Stable and fast connection
  - GPU (optional): For intensive image/OCR processing

Required Software

Operating System:
  - Windows: 10, 11, Server 2016+
  - Linux: Ubuntu 20.04+, CentOS 8+, RHEL 8+
  - macOS: 11+ (Big Sur or higher)

Dependencies:
  - Python 3.8+ (included in installer)
  - .NET Framework 4.8+ (Windows)
  - Updated SSL certificates

Recommended:
  - Antivirus with robot exception
  - Firewall configured to allow communication
  - RDP/SSH server for remote management

Connectivity

Network Requirements:
  - Stable internet connection
  - Minimum bandwidth: 1 Mbps
  - Recommended bandwidth: 10+ Mbps
  - Latency < 200ms to Heptora servers
  - Ports: HTTPS (443) outbound

Firewall Configuration:
  - Allow outbound connections to *.heptora.com
  - Allow websockets (wss://)
  - No inbound ports required

Deployment Topologies

Topology 1: Single Robot on Server

                    ┌─────────────┐
                    │ Heptora     │
                    │ Cloud       │
                    └──────┬──────┘
                           │ HTTPS
                    ┌──────▼──────┐
                    │   Robot     │
                    │   Server    │
                    │             │
                    │ - Processes │
                    │ - Secrets   │
                    │ - Variables │
                    └──────┬──────┘
                           │
          ┌────────────────┼────────────────┐
          │                │                │
    ┌─────▼─────┐   ┌─────▼─────┐   ┌─────▼─────┐
    │Local ERP  │   │Local DB   │   │  Files    │
    └───────────┘   └───────────┘   └───────────┘

Advantages:

Simple configuration
Centralized maintenance
Single point for secrets management

Ideal for:

Small-medium companies
Centralized processes
Low execution volume

Topology 2: Multiple Specialized Robots

                    ┌─────────────┐
                    │ Heptora     │
                    │ Cloud       │
                    └──────┬──────┘
                           │
        ┌──────────────────┼──────────────────┐
        │                  │                  │
    ┌───▼────┐        ┌────▼───┐        ┌────▼───┐
    │ Robot  │        │ Robot  │        │ Robot  │
    │ Office │        │ Server │        │  Web   │
    │        │        │        │        │        │
    │Desktop │        │Backend │        │Scraping│
    │Tasks   │        │Process │        │        │
    └───┬────┘        └────┬───┘        └────┬───┘
        │                  │                  │
   ┌────▼────┐        ┌────▼────┐       ┌────▼────┐
   │ Office  │        │   ERP   │       │Internet │
   │  Apps   │        │   DB    │       │  Sites  │
   └─────────┘        └─────────┘       └─────────┘

Advantages:

Specialization by task type
Better performance (parallelization)
Environment isolation
Horizontal scalability

Ideal for:

Medium-large companies
High volume processing
Different security levels

Topology 3: High Availability

                    ┌─────────────┐
                    │ Heptora     │
                    │ Cloud       │
                    └──────┬──────┘
                           │
        ┌──────────────────┼──────────────────┐
        │                                     │
    ┌───▼────┐                            ┌───▼────┐
    │ Robot  │                            │ Robot  │
    │Primary │◄───────Heartbeat──────────►│Standby │
    │        │                            │        │
    │Active  │                            │Passive │
    └───┬────┘                            └───┬────┘
        │                                     │
        └────────────┬────────────────────────┘
                     │
              ┌──────▼──────┐
              │  Critical   │
              │  Systems    │
              └─────────────┘

Advantages:

Fault tolerance
No downtime
Automatic failover

Ideal for:

24/7 critical processes
Large companies
Strict SLA requirements

Deployment Security

Security Principles

1. Principle of Least Privilege

Robot Service Account:
  ✓ Permissions: Read in input folders
  ✓ Permissions: Write in output folders
  ✓ Permissions: Execute necessary applications
  ✗ NO: Administrator permissions
  ✗ NO: Access to unrelated folders
  ✗ NO: Ability to install software

2. Network Isolation

Secure Network Configuration:
  - Robot in specific VLAN
  - Restrictive firewall rules
  - Only necessary communication allowed
  - Anomalous traffic monitoring

3. Credential Rotation

Rotation Policy:
  - Service credentials: Every 90 days
  - Certificates: Renew 30 days before expiration
  - API tokens: According to provider policy
  - User passwords: According to corporate policy

Robot handles rotation:
  - Alerts before expiration
  - Update process without downtime
  - Post-rotation validation

4. Auditing and Logging

Local Logs:
  - All robot actions
  - Secret access (without revealing content)
  - Process executions
  - Errors and exceptions
  - Cloud communication (metadata)

Retention:
  - 30 days: Detailed local logs
  - 90 days: Aggregated logs in cloud
  - 1 year: Critical audit logs

Monitoring and Observability

Cloud Dashboard

Monitoring View:

Robots (4 active):
  ┌─────────────────────────────────┐
  │ ✓ Robot-Prod-01   │ Active      │
  │   - CPU: 45%      │ 3 processes │
  │   - RAM: 6.2/16GB │             │
  │   - Last ping: 5s ago           │
  └─────────────────────────────────┘

  ┌─────────────────────────────────┐
  │ ✓ Robot-Office-01 │ Active      │
  │   - CPU: 12%      │ 1 process   │
  │   - RAM: 2.1/8GB  │             │
  │   - Last ping: 3s ago           │
  └─────────────────────────────────┘

  ┌─────────────────────────────────┐
  │ ⚠ Robot-Dev-01    │ Inactive    │
  │   - Last ping: 2h ago           │
  │   - Status: Possible restart    │
  └─────────────────────────────────┘

Active Executions (5):
  - FACe Invoicing: 45% (23/50 invoices)
  - ERP Extraction: Starting...
  - Monthly Report: 80% (generating PDF)
  - Stock Sync: Queued
  - Notifications: Queued

Metrics and Alerts

Available Metrics:

Robot Level:
  - Status (active/inactive)
  - CPU, RAM, Disk
  - Heartbeat latency
  - Number of running processes
  - Pending task queue

Process Level:
  - Execution time
  - Success/failure rate
  - Throughput (items/hour)
  - Average duration
  - Frequent errors

Configurable Alerts:
  ⚠ Robot inactive > 5 minutes
  ⚠ Process failed 3 consecutive times
  ⚠ Execution time > 2× average
  ⚠ Task queue > 100 pending
  ⚠ Disk usage > 90%

Frequently Asked Questions

What happens if I lose internet connection?

The robot can continue executing processes that were already in progress. Scheduled processes that should have started during disconnection will execute when connection is restored. Processed data is stored securely locally.

Can I have robots in multiple geographic locations?

Yes, you can have robots in different offices, countries or regions. All are managed from the same cloud panel, but each processes data locally in its location.

Do local robots need constant internet access?

They need internet connection to communicate with the cloud orchestrator (receive tasks, report status), but process execution itself is local and doesn’t depend on internet.

What happens with detailed logs?

Detailed logs with business information are stored locally. Only aggregated logs and metadata (without sensitive data) are sent to the cloud for dashboards and alerts.

Can I migrate from cloud to local or vice versa?

Heptora’s architecture is already hybrid from the start. No migration is necessary; you always have local processing with cloud orchestration. What you can adjust is which processes execute on which robots.

How does Heptora handle backups?

Backups of processed data are your responsibility (they’re in your infrastructure). Heptora maintains backup of process definitions, configurations and metadata in the cloud.

Can I use my own cloud storage?

Yes, you can configure robots to save results to your own Azure Blob Storage, AWS S3, etc. The robot accesses these services directly; Heptora cloud doesn’t see the data.

What security certifications does Heptora have?

The cloud component complies with modern cloud security standards. For sensitive data, the key advantage is that it remains in your certified infrastructure, not Heptora’s.

Can I audit what data is sent to the cloud?

Yes, the local robot has complete logging of all cloud communication. You can review exactly what metadata is transmitted.

Need More Help?

If this guide didn’t solve your problem or you found an error in the documentation:

Technical support: help@heptora.com
Architecture consulting: To design your optimal deployment
Security team: For specific security assessments
Describe your current infrastructure
Explain your security and compliance requirements
Indicate approximate data volumes

Our architecture team will help you design the ideal deployment for your case.

Robot Installation - Local robot installation guide
Secrets Management - How to configure secrets in local robot
Process Templates - Ready-to-use automations
AI Assistants - AI capabilities in the cloud
Security Guide - Security best practices (coming soon)
Enterprise Architecture - Deployments for large organizations (coming soon)

Hybrid Architecture

Overview

Why a Hybrid Architecture?

The Two Components

Cloud Component: The Web Platform

Cloud Component Responsibilities

What’s NOT in the Cloud

Local Component: The Execution Robot

Local Component Responsibilities

Robot Deployment

Complete Workflow

Example: Automatic Invoice Processing

What Traveled to the Cloud?

Hybrid Architecture Advantages

1. Privacy and Security

2. Optimized Performance

3. Flexibility and Control

4. Optimized Cost

Ideal Use Cases

1. Sensitive Information Processing

2. Integration with Legacy Systems

3. High Data Volume

4. Low Latency Requirements

Deployment Considerations

Local Robot Requirements

Minimum Hardware

Required Software

Connectivity

Deployment Topologies

Topology 1: Single Robot on Server

Topology 2: Multiple Specialized Robots

Topology 3: High Availability

Deployment Security

Security Principles

Monitoring and Observability

Cloud Dashboard

Metrics and Alerts

Frequently Asked Questions

What happens if I lose internet connection?

Can I have robots in multiple geographic locations?

Do local robots need constant internet access?

What happens with detailed logs?

Can I migrate from cloud to local or vice versa?

How does Heptora handle backups?

Can I use my own cloud storage?

What security certifications does Heptora have?

Can I audit what data is sent to the cloud?

Need More Help?

Related Resources