Frequently Asked Question
Deploying our technical stack onto client-owned hardware grants a License to Use, not a transfer of asset ownership.
- No Code Manipulation: End users are strictly prohibited from reverse-engineering Docker containers, modifying source code, altering database schemas, or exporting n8n workflow blueprints for external use.
- No Reselling or Distribution: The client cannot copy, clone, distribute, sublease, or resell our technical stack, workflows, or architectural designs to any third party.
- Configuration is Not Creation: Configuring folder routing, editing user-level variables, or tweaking localized prompts does not grant the client any IP rights over the overall software system.
3.1 Perpetual Licensing, Software Updates, and Annual Maintenance Contracts (AMC)
For systems deployed directly on client-owned infrastructure (On-Premises, AWS, or GCP) under a perpetual license agreement, the long-term operation of the software is governed by the following operational boundaries:
- The Right to Run (Perpetual): A perpetual license grants the client the permanent legal right to execute and run the specific version of the software stack (including Docker containers, SQL/Postgres schemas, and n8n workflows) that was delivered to them. This right does not expire, even if active service contracts end.
- The Necessity of an AMC: Access to ongoing technical support, system troubleshooting, security patches, feature upgrades, and n8n workflow optimizations requires a valid, active Annual Maintenance Contract (AMC) or active subscription.
- Impact of No AMC: In the absence of a valid AMC, the system is strictly frozen in its current state. We will cut off all access to upstream update repositories, new Docker image layers, and developer support. The client assumes 100% of the operational and security risks of running unmaintained software.
- Revocation Exception: The license to run the software forever is only revoked if the client violates core intellectual property terms (such as reverse-engineering, reselling, or illegally copying the software stack to unlicensed hardware).
3.2 Lapsed AMC Renewal Policies and Back-Payment Penalties
If a client allows their Annual Maintenance Contract (AMC) to lapse and later wishes to reactivate support and software updates; the following rules apply:
- Reactivation Fees: Reactivating an expired AMC requires the client to pay all accrued maintenance fees retroactively for the period during which the AMC was lapsed, plus a standard reactivation penalty.
- Version Catch-Up Requirement: Because software layers (Docker images, database structures, and n8n nodes) evolve continuously, we will not deploy patches to an outdated system. The client must pay for the mandatory consulting hours required to audit, clear out technical debt, and upgrade their local infrastructure to the current production version.
3.3 AI Large Language Model (LLM) Obsolescence, Retirement, and Migration
The AI landscape changes rapidly, and third-party model providers (e.g., Google, OpenAI, Anthropic) regularly deprecate and shut down older AI models.
- Outside the Scope of AMC: An active AMC covers standard software maintenance, bug fixes for existing logic, and routine platform updates. It does not cover major infrastructure overhauls caused by vendor model retirement.
- Mandatory Migration Projects: When a foundational AI model is retired by the provider (for example, upgrading from Gemini 2.5 to Gemini 3.1 or Gemini 3.5), it changes the underlying API behaviors, prompt execution logic, and token structures. Migrating the system to a newer generation model requires dedicated developer hours to rebuild and test workflows.
- Separate Billable Services: Any work required to transition a client's system to a new AI model generation will be scoped, quoted, and charged separately as a professional services project. If a client refuses to fund the migration project before the original model is retired by the vendor, the AI components of their local deployment will permanently cease to function once the legacy model goes offline.