Limitations
What TVCRpro is, what it is not, and what is honestly true about its stage of maturity. This page is here so the list never has to be re-asked.
What TVCRpro is not
- Not a vendor-review or third-party-risk-management tool.
- Not a productivity suite.
- Not a chatbot.
- Not an AI training pipeline.
- Not a content moderation system.
- Not a security scanner.
What it has not done yet
TVCRpro is pre-pilot. There are no production deployments. Synthetic validation has been performed and is labeled as such anywhere it is referenced. Capability claims are design-intent, not benchmarked outcomes, and are flagged with the chip below wherever they appear:
Pre-pilot, design-intent.
No pilot results are claimed on this site. When pilot results exist, they will appear with their context, their methodology, and the conditions under which they were obtained.
What it depends on
TVCRpro is built on a small, named set of providers. Honest dependency disclosure: if any of these fail, the corresponding TVCRpro capability is at risk.
- Cloudflare — DNS, edge, the Light Analyzer Worker.
- Proton AG — email and the custom domain.
- Microsoft 365 (Canada tenant) — workspace documents, Power Automate workflows, Power BI.
- Adobe Acrobat Sign — e-signature for NDAs and license agreements.
- USPTO — the patent application of record.
See Security for the full sub-processor list and verification paths.
What it cannot do for you
- It does not measure model quality. TVCRpro measures the quality of the interaction with whatever model you use. Switching models will not change a TVCR score by itself; only changing the inputs, the outputs, or the efficiency of the interaction will.
- It does not replace human judgment. A coaching tip is a tip. The decision is yours.
- It does not generate coaching automatically without a scored interaction. An input must be scored first. There is no passive coaching layer.
What it does not yet have
- ISO 27001 certification — not pursued.
- SOC 2 — not pursued.
- Pilot data — none yet, by design.
- A third-party security audit — not yet commissioned.
These are deliberate omissions at this stage, not oversights. They become material when a real licensee prospect requires them and a company exists to commission them.
What may change
The non-provisional patent filing deadline is April 21, 2027. Until that date, the dimensions, the names, and the architecture remain available for refinement under patent counsel review. Production rubric weights remain trade-secret regardless of any naming change.
Any change to the eleven dimension names will be reconciled across the entire site, the Analyzer, and the API in a single coordinated update.