AI Transparency Notice

1. What AI does in UsageX

UsageX Maximize is a ChatGPT plugin that can read and search a repository, plan work, write code, and pass a focused prompt to OpenAI Codex. Codex executes, tests, and reports back to ChatGPT. The supported environment is OpenAI only.

UsageX does not alter, reset, bypass, or circumvent Codex limits. We let ChatGPT and Codex talk to each other for better quality. Any usage saving is a consequence of removing redundant work, not an attempt to get around provider limits. Results vary by task, account, and plugin use. ChatGPT and Codex remain usable as normal at all times.

2. Information sent and stored

To complete a ChatGPT-app request, UsageX may relay the task instructions and only the bounded repository context, file excerpts, prepared edits, verification requests, and execution results needed for that operation between ChatGPT and your paired computer. The repository itself remains on the linked device and is not mirrored to UsageX cloud infrastructure. Unclaimed relay requests normally expire after 10 minutes. Claimed requests can remain until completion or reconciliation; terminal relay and task/run records are eligible for automated deletion after approximately 30 days, with active or recoverable records preserved until resolved. Relevant content may also be processed by OpenAI under the account, contract, settings, and data controls that apply to you.

Operational telemetry uses a pseudonymous identifier and records limited metrics such as provider/model, routing, timing, token counts, errors, plugins, and task length. It is designed not to contain prompt or clarification text, code, files, chat messages, or generated outcomes. Pseudonymous is not the same as anonymous, so this limited metadata can still be personal data. Do not submit secrets or data you are not authorised to process.

3. Plugins

When a plugin is used, UsageX aims to display that plugin in the chat interface. Plugin connections are established and managed through supported OpenAI/Codex integrations. This can use Codex-connected resources, meaning plugin-heavy work may deliver less subscription saving than a task completed without plugins.

A plugin can expose selected information to the connected third-party service. Review its name, requested permissions, terms, and privacy notice before proceeding. Disconnect plugins you no longer use.

4. Training and improvement

ERGO CODE does not train its own public foundation model on your private prompts, files, or outcomes and does not authorise such training without an explicit opt-in. OpenAI's handling of submitted content depends on its applicable plan, contract, settings, and data controls. Review those controls for your account.

We may use pseudonymous, privacy-sanitised operational data to improve reliability, safety, and efficiency. Access and use are limited to those purposes as described in the Privacy Policy.

5. Limitations and human oversight

• Outputs may contain factual, logical, legal, licensing, privacy, or security errors.
• Generated code may be vulnerable, incompatible, or unsuitable for production.
• Similar prompts can produce different results, and outputs may resemble third-party material.
• Savings, performance, and availability are not guaranteed.

You decide whether to accept, edit, test, publish, or deploy an output. Use qualified human review and appropriate testing for material decisions. UsageX is not intended to make legal or similarly significant decisions about individuals and must not be the sole decision-maker in safety-critical, medical, legal, financial, employment, credit, insurance, law-enforcement, or essential-service contexts.

6. Questions and reporting

Report an unsafe, misleading, infringing, or privacy-impacting result to [email protected]. Include enough context for us to investigate, but do not email passwords, API keys, or unnecessary personal data.