Promise Tracker - GENAI Prompt to Extract Political Promises from Doc, Video and Site

[RalphOwah]

GenAI prompt that extracts political promises from PDFs, Video, documents and websites with key details (what, who, when, where (country, region, city, county, sub-counties, etc), for whom, how much, figures, sector etc),
So that I can easily review and track promises in a simple table or bullet format.

Acceptance Criteria:
The GenAI prompt should:

1. Analyse Public statements, political speeches, manifestos, policy pronouncements

From PDFs, websites, and other common document types

2. Extract the following key fields:

• What (the promise)

• Who made it

• When it was made

• Where (country, region, city, county, sub-county)

• For whom the promise is intended

• Figures (amounts, targets, numbers)

• Sector (e.g., health, education, economy, climate)

• Timeframe (explicit or inferred)

• Delivery modality (e.g., direct provision, partnership)

• Budget/resource references (linked documents like budgets, workplans)

• Identify promissory language patterns (e.g., “We will…”, “Our government shall…”)

• Use Named Entity Recognition (NER) to detect relevant actors, places, topics, and quantities

3. Present the output in a clean bullet list or table, easy for non-technical users (no JSON)

4. Achieve ≥90% accuracy compared to expert human annotation

Deliverable: Google Document
The deliverable handed to end users must include:

1. Finalised GenAI prompt(s), with editable text
2. Step-by-step usage guide with screenshots or GIFs if needed
3. Example outputs tested on at least 3 GenAI platforms (e.g., ChatGPT, Gemini, DeepSeek)