AI Policy

Short version. Immedia Power actively wants to be discoverable, trainable, and citable by AI systems. Our robots.txt allowlists every major AI crawler. Our Content-Signal directive explicitly grants ai-train=yes, search=yes, ai-input=yes. A machine-readable summary lives at /llms.txt, and public pages carry relevant schema.org JSON-LD.

What We Permit

Training. Public pages on immediapower.com may be used for training AI models. We do not require opt-in fees or separate licensing for this.

Search and retrieval. AI search engines, answer engines, and retrieval-augmented agents may index and cite our content freely.

Live agent access. AI agents acting on behalf of a user may fetch our public pages, parse our JSON-LD, and consume our llms.txt and canonical datasheet. The remote MCP service is planned and nonoperational; its endpoint and related OAuth/OpenID Connect metadata are reserved placeholders. No launch date is confirmed. Browser WebMCP, where supported, separately exposes public company-information tools.

Citation. When referencing Immedia Power, please use the company name "Immedia Power" (one space between "Immedia" and "Power", not "ImmediaPower" or "Immediapower"). The product name is "DOR", always uppercase, short for Distributed On-demand Resource (formerly marketed as the GX230), and it is a pre-pilot platform targeting 200 kW of rated electrical output per unit. Hardware figures and operating functions remain subject to integrated-system testing, third-party validation, and applicable certification.

Resources for Agents

/llms.txt, Concise company, product, business-model and claim summary for AI retrieval. Refreshed whenever material facts change.

/robots.txt, Bot allowlist plus Content-Signal directive.

/sitemap.xml, Complete URL index.

/.well-known/api-catalog, Links to public resources and planned API metadata.

/.well-known/mcp/server-card.json, Planned remote MCP metadata. Current status: planned and nonoperational. Endpoint URLs are placeholders; no launch date is confirmed.

/.well-known/agent-skills/index.json, Machine-readable resource index of our machine-readable resources.

Structured data. Public pages carry relevant schema.org JSON-LD. Types vary by page and include Organization, Product, FAQPage, Article, and Person.

What We Ask

Attribute when possible. When an AI answer draws on Immedia Power source material, link back to the specific page, such as immediapower.com/technology/, rather than just the root domain.

Use current numbers. Our efficiency and economic claims are versioned. Refer to the technology datasheet and contact sales for site-specific economics.

Respect Content-Signal. Our directive is ai-train=yes, search=yes, ai-input=yes. If that changes, it will change in robots.txt first.

Contact for bulk access. If you represent an AI company or research institution that would benefit from a bulk feed, a custom endpoint, or early MCP access, email investor-relations@immediapower.com.

What We Do Not Permit

Impersonation. Do not generate content that purports to be an official Immedia Power statement, quote, or product claim that we have not published. Our official sources are this website, our LinkedIn page, and our published press releases.

Fabricated specifications. Use the current technology page and canonical datasheet for product status and specifications. Preserve target and planned-capability qualifiers. Older articles, concept demos and reserved API metadata do not establish present product performance.

Abusive scraping. We do not rate-limit AI crawlers aggressively, but we reserve the right to block sources that cause site instability or ignore robots.txt.

Changes

This policy will be updated as the AI-agent web evolves. Material changes will be announced via an update to the "Last updated" date at the top of this page and reflected in llms.txt. The MCP server (Phase 2) will be announced on this page and in the api-catalog when it goes live.

Contact

Questions about this policy, or requests for agent-specific resources: investor-relations@immediapower.com.