ANATYLICA
Request accessLogin

AI Use & Limitations

Last updated: 13 April 2026

Some ANATYLICA features may use machine-learning, language-model, or other automated techniques to structure, summarize, classify, rank, or generate outputs. This statement explains the limits of those features and how they should be interpreted.

How AI may be used

  • Summarizing text, documents, or data points.
  • Generating labels, signals, or classifications.
  • Supporting research and analytical workflows.
  • Providing draft responses, insights, or organizational structure for internal review.

Key limitations

  • AI-generated or model-assisted outputs may be inaccurate, incomplete, biased, stale, or misleading.
  • Outputs may incorrectly infer relationships, materiality, sentiment, relevance, chronology, or confidence.
  • Models may overstate certainty or generate plausible but unsupported statements.
  • Even when outputs look polished, they should not be treated as verified fact without independent checking.

Human responsibility

Automated outputs are intended to support professional workflows, not replace human judgment, legal review, compliance review, or investment decision-making. You remain solely responsible for verifying material inputs and outputs before relying on them.

ANATYLICA may use confidence scores, validations, review workflows, or other controls, but those controls do not eliminate the possibility of error.

Training and product use

Unless a separate contract states otherwise, ANATYLICA should disclose clearly whether customer content, customer prompts, or customer outputs are used to train internal or external models. If ANATYLICA intends to use customer-controlled data for model training or tuning, that use should be expressly documented in the governing customer contract and privacy documentation.

At this stage, ANATYLICA should default to a conservative position: customer-controlled non-public data is used to provide the service and maintain security and quality, but not for broad third-party model training unless expressly agreed.

Regulatory and operational caution

Model-assisted outputs should not be treated as a substitute for regulated investment advice, legal review, compliance sign-off, accounting treatment, tax advice, or any other regulated or high-stakes professional function.

Related documents: Disclaimer, Terms of Use, and Privacy Notice.