Tool & guide

Delaware Statutory Trust Reviews: Review Claim Verification Queue

By Sequoia ResearchPublished Read as Markdown

Review Claim Verification Queue

Users extract material claims and assign a source, verification owner, status, and unresolved question. The queue supports human checking of both manually written and AI-generated summaries.

For an export, select Download in the tool, then Save file when it appears below. Your entries stay in your browser.

What Is This Article About?

This guide explains a tool for tracking claims that need human review. You enter the exact statement, its source, the person responsible for checking it, and what remains unresolved. The queue works for manually written summaries, AI-generated summaries, and passages taken from a source. It organizes the work but does not verify a claim or approve an investment.

Table of Contents

What the Claim Verification Queue Does

The tool turns a collection of statements into a defined review task. Each row holds a claim, an origin label, a source reference, a review owner, a status, dates, an evidence note, and an unresolved question. You can work with one to 30 claims in a single queue.

The output keeps every claim visible with its recorded conclusion and next step. It also counts open, closed, unassigned, and overdue rows. These counts describe the work you entered. They are not a measure of the quality of a DST, sponsor, broker, or research report.

A claim marked Supported means a person has entered that judgment and the required fields. The tool has not opened the document or judged the reasoning. A complete-looking record can still be wrong. Use the queue to support a careful human process, not to replace it.

FINRA's private-placement guidance discusses checking issuer claims and assessing outside research with reasonable care. Those requirements apply in the member-firm context. This educational queue does not complete that investigation or establish compliance with it. FINRA Regulatory Notice 23-08.

Extract One Checkable Claim at a Time

Start with a statement that can be checked against a record. “The loan has no prepayment charge after year three” is more useful than “good debt.” The first identifies a term and a period. The second is a broad judgment that needs several underlying facts.

Split compound statements when their parts require different evidence. “The sponsor has a long history and this offering has low fees” belongs in at least two rows. A supported history claim cannot close a fee question merely because both appear in the same sentence.

Keep the exact wording you intend to use. If the source says “may,” do not silently turn it into “will.” If a summary leaves out a condition, record the full claim being checked and explain the condition in the reason field. The queue should show why the conclusion is limited.

The claim field allows 1,000 characters. That is room for a focused sentence and its context, not a whole report. If a statement cannot be reviewed without a long chain of assumptions, separate the steps. Name the offering or document in the claim when that is needed to avoid confusion.

Record the Source and the Wording Origin

Choose one of three origin labels: Manual summary, AI-generated summary, or Source extract. The label describes how the wording entered the queue. It does not establish whether the statement is true. A copied source passage can be outdated, misunderstood, or applied to the wrong offering.

The source-reference field should identify the title, date or version, and relevant page or section. Add a short limit if it matters. For example, “loan summary, section 4; charge schedule not included” tells the next reviewer more than “PPM.” The source field allows 800 characters.

An optional public link can point to the source. It must be a complete http or https address without embedded login credentials. The tool does not visit it. An entered link appears in the result for a person to open; its presence does not prove that the page is available or supports the claim.

Use a plain reference for a private document held in an approved secure location. Do not paste account numbers, tax identifiers, or confidential text that the task does not need. The queue is not a secure document vault. A short source label can guide an authorized reviewer to the correct file.

Assign a Human Review Owner

The owner is the person responsible for the next review action. A short name or agreed label is enough. The tool does not verify identity, assign legal authority, or send a message to that person. Agree on responsibility outside the page before relying on the label.

A blank owner stays unassigned. The output calls for an owner to be named and includes the row in the unassigned count. It does not assume that the person who opened the page owns every task. That distinction matters when several people discuss the same research file.

The owner field allows 80 characters. Use it consistently so the results filter can find related work. “Reviewer A” in one row and “A” in another remain different text for filtering purposes. The tool does not maintain a staff directory or merge aliases.

Responsibility is also separate from expertise. A coordinator can gather documents while a legal or tax question needs a qualified professional's judgment. Record that handoff in the unresolved question. Do not use the owner field to imply that someone has approved a subject outside their role.

Choose a Status That Matches the Evidence

There are seven statuses. Not reviewed means the source check has not been recorded. In review means work is underway. Neither status says the claim is reliable or unreliable.

Supported records a human conclusion that the exact claim is supported within the stated source and limits. Partly supported means some of the claim has support but an important part remains open. Contradicted means the reviewer has identified evidence or reasoning that conflicts with the wording being used.

Unresolved keeps a question open when the available work does not settle it. Outside scope records a reason for excluding the claim from this review. It does not establish that the claim is true or false. A scope change may make the row relevant later.

Use the reason field to explain the choice. A bare status cannot tell another reader why a claim was accepted, limited, or rejected. The reason can include a calculation, a condition, or a conflict between sources. It allows 1,200 characters.

The unresolved-question field holds the next evidence request or correction task. It allows 800 characters. Be specific enough that someone can tell when the task is done. “Get the charge schedule that applies to this loan” is clearer than “check debt.”

Meet the Evidence Requirements for Checked Statuses

Supported, Partly supported, and Contradicted rows require a named owner, a human check date, a source reference, and a reason. Partly supported and Contradicted also require an unresolved question or correction task. These requirements prevent an empty status change from looking like a documented review.

Outside scope requires an owner, check date, and reason. It does not require a source reference because a scope decision may concern the purpose of the review itself. Include a source when the reason depends on a document. Unresolved requires an open question, even when its owner or evidence is still missing.

The fields are requirements for a record, not an automated proof test. Typing a document name does not show that it was read. Writing “looks fine” does not make a reason persuasive. The responsible reviewer still has to inspect the source and assess the exact claim.

The SEC's performance bulletin distinguishes actual results from targets and projections and urges readers to examine calculation methods. A well-filled queue row should preserve those distinctions. It should not let a single payment notice become proof of an annual investment return. SEC performance-claims bulletin.

Use Follow-Up Dates Without Inventing Deadlines

The review cutoff is a required calendar date. It tells the tool when to assess follow-up dates. The fictional example uses October 16, 2026. Replace that date when using the queue for real work. The tool does not silently move the cutoff forward each day.

An open row with a follow-up date before the cutoff is Overdue. A date equal to the cutoff is Due on cutoff. A later date is Later. A blank follow-up date remains No date. Supported and Outside scope rows are treated as closed for this workflow, so their date label is Closed.

These are planning dates chosen by the user. They are not legal, tax, contract, or regulatory deadlines. A row marked overdue does not establish that a firm has violated a duty. The underlying task may need a separate professional review of any actual deadline.

Human check dates cannot fall after the selected cutoff. All dates must be valid calendar dates from 1900 through 9999. If you move the cutoff backward, a later recorded check will need attention before the queue can be built. The tool will not pretend that review work occurred earlier than its entered date.

Work Through the Fictional Four-Claim Example

The starting record has four invented claims. All document names, people, dates, and figures are demonstration entries. They do not describe a real offering or review.

The first claim says that one fee equals $300,000 because it is 1% of a $30 million purchase price. Reviewer A marks that narrow calculation Supported and records the source and limit. The arithmetic is correct: $30 million multiplied by 0.01 equals $300,000. The note expressly does not establish total costs.

The second claim comes from an invented AI summary. It says that a $1,500 quarterly payment on $100,000 proves a 6% annual investment return. The quarter's payment is 1.5% of that starting amount; multiplying it by four gives a 6% annualized payment pace. That arithmetic does not prove a full year's payments, future payments, or total return. Reviewer B marks the wording Contradicted and assigns a correction task.

The third claim says a loan can be prepaid without a charge after year three. The supplied fictional excerpt describes a window starting in year four but does not establish that charges end. Reviewer A marks it Partly supported and asks for the relevant schedule and conditions.

The fourth claim describes a 45-day report schedule from a fictional service summary. It remains Not reviewed and has no owner. The queue does not apply that schedule to actual DSTs. The task is to check the exact commitment, scope, exceptions, and version in this example.

At the example cutoff, the counts are four claims, three open, one closed, one unassigned, and one overdue. The third row is due on the cutoff; the second is due later. None of those counts says the investment is good or the research is complete.

Review AI Summaries With the Same Source Discipline

An AI-generated summary can make a long document easier to scan. It can also lose a condition, mix dates, or state an inference as a fact. Treat its source links and calculations as items to check, just as you would with a manually prepared summary.

Open the cited record and find the actual passage. Check the offering name, period, unit, and page. If the summary uses a quotation, confirm that the wording is present and that nearby conditions do not change its meaning. The queue itself performs none of these steps.

Do not assume that manual wording is more reliable merely because its origin label differs. The same status rules apply to all three origins. The AI-origin count is a description of the file, not a risk score. A human still needs to decide which claims matter and whether the supporting work is enough.

Keep a correction specific. Replace a wrong amount, add the missing period, or narrow a statement to what the source supports. If the claim remains unresolved, keep it out of any conclusion that depends on it. A polished paragraph should not hide an open evidence question.

Filter the Queue and Read the Counts

After entering the rows, select Build current queue. The output shows the entire-queue counts and the number of rows visible under the selected filters. You can show all claims, open claims, overdue claims, or a single status. The owner filter matches text contained in the owner field, without regard to letter case.

Filters change the view, not the saved inputs or the export. A result with no matching rows does not mean the queue is empty. The tool says how many rows are shown and keeps the full counts visible. Export always includes every current row.

An open claim is any status other than Supported or Outside scope. Thus Contradicted remains open because the wording or conflict still needs action. A closed count is based on entered status, not independent proof that all work is done. Read the reasons and questions alongside the totals.

Save the Record and Reopen Changed Claims

Any edit to the form clears the old output and disables export. Build the queue again to use the new entries. Removing a claim, clearing the form, or loading the example requires an in-page confirmation. Cancel preserves the entries. Editing while a confirmation is pending cancels that pending action.

The JSON export contains the entire current queue, its source references, human judgments, dates, counts, next steps, limits, and generation time. On Writing, use the tool's Export control and then the Save file button below the frame when shown. Read the saved file to confirm it contains the intended record.

There is no account, automatic storage, import, or version history. Refreshing clears edits. Keep separate dated exports if earlier conclusions matter. The generation time records when the file was made; it does not prove when a source was checked or when its facts were true.

Reopen a claim when its wording or source changes in a meaningful way. A new report may update a payment figure without changing an agreement. A new supplement may alter a term. Record what changed and why the prior conclusion no longer answers the current question.

Understand the Tool's Limits

The queue records what users say. It does not fetch documents, assess authors, detect fabricated citations, read private files, or test whether a source supports a statement. A required field can be filled with poor evidence. The human process remains responsible for that distinction.

FINRA's 2026 oversight report discusses failures to preserve evidence of investigation and resolve concerns about private placements. It was published in December 2025. A queue can help organize records, but it is not the complete work the report discusses or a substitute for professional duties. FINRA oversight report, printed pages 53–55.

Frequently Asked Questions

Does Supported Mean the Tool Verified the Claim?

No. It means a person selected that status and supplied the required record fields. The tool checks that the fields are present, not that the evidence is correct. Read the reason, source, and limits before relying on the conclusion.

Can I Track Both Human and AI-Written Claims?

Yes. Choose the origin that describes how the wording entered the queue. The same review rules apply to every origin. No label establishes reliability, and the tool does not call an AI service or automatically review the text.

Why Does a Contradicted Claim Stay Open?

Its wording or source conflict still needs a correction task. The queue requires that task to be recorded. After a person resolves it, the claim and status can be revised with a new reason and check date. Keep an earlier export if you need a record of the original problem; the tool has no version history.

Does Outside Scope Mean the Claim Is False?

No. It means a reviewer gave a reason for excluding it from this particular review. That is a scope decision, not a truth finding. Reopen the row if the purpose changes or the statement becomes material to a conclusion.

Are Filtered-Out Claims Included in the Export?

Yes. Every current row is exported. Filters only change what is visible in the results. The output also shows full-queue counts so a filtered view does not look like the complete record by mistake.

Will the Tool Save My Work Automatically?

No. Export the current queue before leaving or refreshing the page. This version cannot import the file or restore earlier entries. The downloaded JSON is a record for you to retain and review, not an automatic backup service.

About Sequoia Research

Sequoia Research, LLC provides independent research on DSTs and other real estate investment offerings. This educational tool helps readers assign source checks, preserve evidence limits, and keep unresolved claims visible.

Sources