Workspace

2025.04.1

Use of Artificial Intelligence· Corgi Innovation ETFCORGX

Process

0/2required QA

Workflow: no run linkedStatus: draft

ProgressGates OK

0%All checks pass.

4/4done

Instant rules: pass = Done, fail = Blocked. Running / Pending apply to workflow steps elsewhere.

Submit / approve use the saved body; preview below follows the editor (including unsaved text).

Document body length ≥ 100 characters
Structured body (multiple paragraphs or sections)
Document title available for filing context
Body references AI / automation (slug expects AI disclosure) — Include AI, artificial intelligence, or model language.

2 required checklist items open — finish in Checklist tab.

draft

Introduction

Portfolio management decisions for the Fund remain subject to human judgment by the Adviser’s investment personnel. Automated outputs, scores, or summaries do not by themselves cause purchases or sales of securities for the Fund. Personnel are expected to exercise professional skepticism, corroborate material facts using independent sources where practicable, and document rationales for material decisions in accordance with the Adviser’s policies.

---

Permitted categories of use (illustrative)

Without limitation, the following categories describe representative uses that may apply from time to time: (i) parsing and tagging unstructured text from issuer filings, news, and research vendor feeds to flag topics for analyst follow-up; (ii) constructing factor or risk dashboards that summarize exposures at the portfolio or issuer level; (iii) assisting in the preparation of first drafts of internal memoranda, checklists, and compliance questionnaires, which are reviewed and edited by personnel before reliance; (iv) monitoring operational metrics related to trade settlement fails, corporate action processing, and cash management alerts; and (v) supporting cybersecurity log analysis and anomaly detection in coordination with information technology staff.

---

Data sources and data quality

Models may rely on third-party datasets, vendor estimates, crowdsourced information, and publicly available regulatory filings. Such data may contain errors, omissions, stale values, or biases. The Adviser seeks to use data vendors that maintain contractual representations and, where available, independent verification; however, no assurance can be given that data will be accurate or complete. Use of web-scraped or user-generated content may increase the risk of misinformation or manipulation.

---

Model risk and limitations

Statistical and machine learning models are typically calibrated using historical relationships that may not persist. Overfitting, regime change, structural breaks, and correlation breakdowns may cause model signals to be misleading. NLP and LLM systems may “hallucinate” facts, mis-cite sources, or produce internally inconsistent reasoning. Model outputs should be treated as hypotheses rather than definitive conclusions unless validated through independent procedures.

---

Third-party and cloud services

Certain tools may be hosted by third-party cloud providers or software vendors. Use of such services may involve transmission and temporary storage of data outside the Adviser’s premises. The Adviser maintains vendor diligence procedures, security reviews, and contractual protections designed to address confidentiality, data retention, subprocessors, and incident notification; however, no security program can eliminate all risk of unauthorized access.

---

Confidentiality and material nonpublic information

The Adviser maintains policies addressing insider trading, information barriers, and handling of confidential issuer information. Personnel are trained not to input material nonpublic information into vendor tools that are not approved for confidential data. Breaches of policy may result in disciplinary action and referral to regulators.

---

Recordkeeping and supervision

The Adviser maintains books and records in accordance with applicable Advisers Act and 1940 Act requirements. Supervisory procedures address approval of new tools, periodic review of vendor relationships, and escalation pathways when model behavior appears anomalous. The Chief Compliance Officer oversees testing of policies designed to prevent violations of federal securities laws.

---

No guarantee of outcome

There is no assurance that the use of models or AI-assisted workflows will improve investment performance, reduce risk, or increase operational efficiency. The Adviser may modify or discontinue specific tools without shareholder approval except to the extent such change constitutes a material change requiring an update to the Fund’s registration statement.

---

Shareholder inquiries

This disclosure is a working draft for internal review. Additional detail regarding specific vendors, model classes, and governance testing may be included in the SAI or in responses to Board requests. Shareholders should read the final prospectus and SAI when available.