Private tenant. Audit trail. No banned platforms.
A single-tenant AI stack deployed on Oction-owned infrastructure, trained on BBLC-corrected source records, and gated so no client data crosses into a third-party model or cloud.
AI assistants currently describe BBLC as a cosmetics distributor because the public record is stale. Fixing that requires controlling the source records, not prompting ChatGPT.
Three problems a public SaaS cannot solve. First, the corrected entity graph, benchmark data, and compliance documents are sensitive issuer material; they should not be uploaded to OpenAI, Anthropic, or any US-cloud API. Second, the work is continuous: a one-time schema markup fix will be reverted by data providers unless it is monitored and defended. Third, securities content needs an audit trail; a black-box API gives you none.
A sovereign tenant solves all three. The model and data stay inside an isolated Oction partition on our own nodes. Every query, source citation, and content generation is logged. Retrieval crawlers can read the corrected public record, while confidential drafts never leave the tenant.
This is not an investor-visibility or stock-promotion tool. The tenant is for product, compliance, and corporate-record accuracy. No feature is designed to move the share price, target investors, or generate trading interest.
Each module runs inside the BBLC tenant. They share a single knowledge graph and a unified audit log.
Canonical profiles for BBLC, InfernoGrid, Capistral, Koilink, and Orvexa. Ticker, exchange, legal names, operating descriptions, and the corrected sameAs graph in one controlled source.
Ingest press releases, filings, wire copy, and counsel-approved FAQs. De-duplicate, version, and flag stale or conflicting claims before they reach any public channel.
schema.org Corporation markup, robots.txt crawler policy, source-record correction playbooks, and a monthly scored baseline across ChatGPT, Claude, Perplexity, and Bing Copilot.
Product documentation, benchmark reports, Capistral issuer guides, and X/Reddit drafts generated from approved source material, not from a public model trained on competitor content.
Every retrieval, generation, and user action is logged with timestamp, user, source documents, and output hash. Exportable for counsel review and regulator inquiry.
Single-tenant deployment. BBLC data lives in its own partition on Oction-owned Mac Mini nodes, connected via Tailscale mesh. No client data is commingled with other Oction accounts.
No banned platforms. We do not use OpenAI, GitHub, Vercel, or Supabase. Models are served through the Oction LiteLLM proxy from approved providers; embeddings and vector storage are self-hosted.
Anonymize before storage. Any user or customer data is de-identified and k-anonymity gated before it enters the knowledge base. Per-tenant isolation is enforced at the filesystem, database, and network layers.
Deployment target. Go-live within 21 days of signature and receipt of source materials. Initial access via secure web portal; optional VPN-only mode on request.
| Item | Detail | Amount |
|---|---|---|
| Initial deployment and configuration | Tenant provisioning, model routing, knowledge-base seeding, schema markup, crawler policy, audit logging, user access controls, go-live | $25,000 |
| Monthly platform and support | Model access, storage, compute, monitoring, monthly source-record refresh, scored AI-search baseline report, support | $10,000 / month |
| Minimum term | Three months, then month-to-month | 90 days |
| Total to go-live | $35,000 |
Payment. 50% of the initial deployment fee on signature; 50% on go-live. Monthly fees payable in advance. Increases in compute or storage above the included tier are billed at cost with prior approval.
Scope boundaries. Included: the five capabilities above, up to five named BBLC users, and 50 GB of indexed documents. Excluded: securities counsel review, PCAOB audit support, OTC Markets fees, third-party wire distribution, photography or video licensing, and translation.
This deployment complements the Oction Consulting and Marketing Inc. (Oction Agency) channel program. It gives BBLC the controlled source of truth that makes the marketing spend defensible: accurate AI search, auditable content generation, and a compliance document pipeline.
Next step: confirm scope and provide source materials (corporate records, approved fact sheets, existing deck and release library). We provision the tenant and begin the knowledge-base build the following business day.