# Prompt-to-brief pack

Self-initiated demonstration · Synthetic data

Three original templates. Examples are synthetic. Prompt assembly is local text formatting; no AI model executes here. Illustrative outputs are stored examples, not live responses or measured results.

# Research brief

# Research brief

## Required input
Research question, audience, supplied source extracts with dates, and decision to support.

## Supplied input
Audience: a studio operations lead. Question: which delivery format should the team review? Source A, sample brief dated 2026-09-15: reviewers need editable rows. Source B, sample meeting note dated 2026-09-16: a written issue log is required. Both sources are synthetic.

## Constraints
Use supplied evidence only. Distinguish observations from conclusions. Link each factual claim to a supplied source. Preserve uncertainty and conflicting findings.

## Output format
Decision context; source register; supported findings; uncertain findings; next questions.

## Quality review
- Every finding has a supplied evidence reference
- Dates and gaps are visible
- Conclusions do not exceed the evidence

## Limitations
The template does not browse, verify sources, or produce a live model response. A person must review evidence, permissions and claims.

## Illustrative output
ILLUSTRATIVE OUTPUT — not a live AI response.
Observation: the sample brief requests editable rows [A]. The sample meeting note requests an issue log [B].
Recommendation: prepare an editable table with a separate issue log; confirm the file format with the reviewer.
Unknown: required import schema and sign-off date.

## Before use
Review permissions, replace the sample input, and inspect generated results against every quality criterion.

---

# Structured data extraction

# Structured data extraction

## Required input
Authorized source text, required fields, output schema, and missing-value convention.

## Supplied input
Synthetic source: Alderwick Studio, North region, requests a research summary. Brackenmere Systems requests spreadsheet cleanup; region not supplied.

## Constraints
Extract only explicit values. Keep missing fields null. Do not infer contact details, locations or dates. Preserve the source snippet for each record. Flag conflicting values.

## Output format
JSON array with organization, region, requested_service, source_excerpt and review_issues.

## Quality review
- Output follows the supplied schema
- Missing values remain null
- Each value is traceable to its source text

## Limitations
Formatting checks are not extraction-accuracy measurements. Real files can have ambiguous layouts, OCR errors, omissions and restricted data.

## Illustrative output
ILLUSTRATIVE OUTPUT — not a live AI response.
[{"organization":"Alderwick Studio","region":"North","requested_service":"research summary","source_excerpt":"Alderwick Studio, North region, requests a research summary.","review_issues":[]},{"organization":"Brackenmere Systems","region":null,"requested_service":"spreadsheet cleanup","source_excerpt":"Brackenmere Systems requests spreadsheet cleanup; region not supplied.","review_issues":["Region missing"]}]

## Before use
Review permissions, replace the sample input, and inspect generated results against every quality criterion.

---

# Content outline

# Content outline

## Required input
Audience, purpose, approved facts, desired tone, required sections, and exclusions.

## Supplied input
Audience: a small team receiving a cleaned contact file. Purpose: explain how to review it. Approved facts: duplicates are grouped by exact normalized company and email; missing fields stay blank; an issue log is included. Tone: practical and concise.

## Constraints
Use plain English and supplied facts. Do not add invented benefits, results, prices or testimonials. Mark unanswered questions instead of filling them.

## Output format
Working title; audience need; ordered sections with factual notes; unanswered questions; review checklist.

## Quality review
- Outline addresses the stated audience and purpose
- Unsupported claims are excluded
- Open questions remain explicit

## Limitations
This is an outline template, not factual verification, legal advice, a publishing service or a live AI integration.

## Illustrative output
ILLUSTRATIVE OUTPUT — not a live AI response.
Working title: Reviewing your cleaned contact file
1. Open the reviewed output and preserve the original.
2. Inspect exact-match duplicate groups.
3. Review missing fields and the issue log.
4. Confirm unresolved decisions before import.
Open questions: target CRM schema and who approves the import.

## Before use
Review permissions, replace the sample input, and inspect generated results against every quality criterion.