How To Use AI As Analyst

Upload your SOW

Upload the SOW you built manually or in Scope Builder (PDF, DOC, or DOCX). AI As Analyst automatically performs a formatting validation scan to ensure your SOW meets the required SOW structure (see How To Use Scope Builder for details on SOW formatting requirements), and provides a specific, itemized list of any syntax that requires modification.

Review your cost estimate — before anything is generated

Once your SOW passes validation, AI As Analyst runs a section-by-section cost estimate and shows you the full breakdown, then asks you to confirm: Total estimated cost: $X.XX per competitor report. Would you like to proceed? (y/n). Nothing is researched until you say yes.

Confirm your primary competitor

AI As Analyst extracts the header fields from your SOW (Project Name, Geographic Focus, Competitor, Product) and — before committing any research budget — runs a live verification check to confirm the named competitor actually exists and is realistically researchable, with a clear yes/no and the reasoning behind it.

Add or remove additional competitors

Want to research more than one competitor against the same SOW? Enter each one as competitor name | product name | geography (use Shift-Enter to add additional competitors). Each addition is verified the same way as your primary competitor before it's added to the batch. Type remove [name] to drop one, or press Enter with nothing typed to finish.

Confirm the batch and launch

You'll see the full competitor list and total batch cost one more time before launch. Once confirmed, AI As Analyst creates a fully independent research instance/job for each competitor — these run as parallel, isolated processes, not a queue, so a five-competitor batch doesn't take five times as long. Each research job will be listed on your My Workspace page so you can see it "Running" and when it is "Completed", after which you can download the "Results"

Governed, section-by-section research

Each instance works through your SOW one section at a time, researching every question with live web search — all run under the same shared proprietary governance standards against your SOW's exact question text, pulled verbatim rather than re-interpreted.

Receive your report

Once every section is researched, AI As Analyst combines all of it into a single report — formatted directly against your original SOW, so the final output mirrors the exact structure and questions you scoped. Once completed, you can download the report results by clicking the "Results" button for the report shown on your My Workspace page.

Wrap every answer in RESEARCH FINDINGS / END RESEARCH FINDINGS

Every answered question's evidence needs a literal, single-spaced RESEARCH FINDINGS opener and END RESEARCH FINDINGS closer. Each opener needs exactly one closer before the next question's opener begins. AI TruthCheck™ blocks the submission outright if no openers exist anywhere, and blocks it again if openers and closers don't match up one-to-one or aren't correctly paired — in both cases you'll get a specific error naming the mismatch.

Never place content directly after a closer on the same line

Citations, reference links, or footnotes placed right after END RESEARCH FINDINGS on the same line will be flagged and blocked — AI TruthCheck™ quotes the offending line back to you. Put anything like that in its own clearly separated block instead.

Use a colon after every question identifier — never a period

Every question restatement must read exactly "ID: question text" — for example A5: A description of.... Using A5. instead of A5: is blocked with an error naming the exact identifier and line.

Main bullets — stick to •

Use a real bullet character for every main bullet question — either the standard Unicode bullet (•, U+2022) or Word's Wingdings-style bullet (\uf0b7).

Sub-bullets — plain hyphen and space only

Sub-bullets must start with "- " (hyphen, space) — never an asterisk ("* "), and never an en-dash or em-dash.

No literal markdown symbols — outside of RESEARCH FINDINGS

Typed **bold** or __underline__ characters except those that lie between RESEARCH FINDINGS and END RESEARCH FINDINGS are not permitted anywhere in the document.

Structural labels must be exact, undecorated text

Project Description, Background Notes, Section Context, and the metadata prefixes (Project Name:, Language Discipline:, Geographic Focus:, Competitor:, Product:) must appear as plain, case-sensitive, standalone text. A decorated or malformed label (e.g. **Section Context**) is not permitted.

Section headers must sit directly above their first bullet

A line is only recognized as a section header if the very next substantive line is itself a bullet. A header-like line with anything else in between might be flagged as a likely mislabeled header.

Keep page footers out of the body text

Page-number stamps like "47 | Page" should be a true visual PDF/Word footer, not text embedded in the readable content layer.