Library — Prompts, Knowledge, Skills, Connectors

Content → monetization plan

Turn a piece of sports content into ad placements, native units, and a betting/DFS layer with compliance flags.

You are a sports monetization strategist. Given the article below, return: (1) recommended ad units and placements with a yield rationale, (2) one native sponsored idea, (3) any betting/DFS unit IF appropriate, with the required age/state gates and disclaimers. Refuse betting units on injury or youth content.

ARTICLE: """{paste}"""

Recap → social pack

One game recap becomes five platform-ready posts.

Turn this recap into 5 social posts: an X post (<280 chars), an Instagram caption with 3 hashtags, a TikTok hook line, a LinkedIn angle, and a push-notification line. Keep the team's voice. No betting or predictions.

RECAP: """{paste}"""

Responsible-gambling promo copy

Promo copy that ships with the guardrails built in.

Draft betting-promo copy for the offer below. Requirements: include "21+, gambling problem? Call 1-800-GAMBLER", note eligibility varies by state, never imply guaranteed wins, never target minors or self-excluded users. If the context is unsafe (injury, youth), refuse and say why.

OFFER: """{paste}"""

Fan FAQ from data

Given a team's live data, generate a grounded fan FAQ.

Using ONLY the DATA below, write 6 FAQ-style Q&A pairs a fan would ask this week. Cite the data for each answer. If something isn't in the data, omit it. No predictions or betting advice.

DATA: """{paste schedule/standings JSON}"""

Sponsor one-pager

A brief becomes a sponsorship one-pager.

From the brief below, write a one-page sponsorship pitch: audience snapshot, 3 activation ideas, a simple package table (tier / deliverables / est. value range), and a clear next step. Keep numbers as honest ranges, not invented precision.

BRIEF: """{paste}"""

Trivia generator

Stats data becomes verifiable trivia questions.

From the DATA below, write {N} multiple-choice NFL trivia questions. Each correct answer must be verifiable from the DATA; the wrong options must be plausible. Return JSON: [{question, options[4], answer, explanation}]. No betting or predictions.

DATA: """{paste}"""

RAG vs fine-tuning for sports data

Sports facts change by the minute, so they belong in retrieval, not model weights. Use RAG to ground a strong base model in live data; use fine-tuning to shape behavior (tone, format, refusal), never to memorize facts. Conflating the two is the most common, most expensive mistake.

Responsible-gambling compliance, the basics

Age and state-eligibility gates, required disclaimers (1-800-GAMBLER), no targeting minors or self-excluded users, and no betting units against injury or youth content. For a publisher this isn't red tape, it's the safety layer that makes betting revenue usable next to editorial.

Why native video drives ad revenue

Native in-line video sits inside the content and reads as part of the page, so it earns higher-value demand than saturated display. It's the lever behind real YoY ad-revenue growth at sports publishers, and why "where and how the unit is placed" beats "more units."

The sports data API landscape

Free/prototype: ESPN's public endpoints, balldontlie, TheSportsDB, no commercial rights. Licensed/real: Sportradar, Stats Perform, SportsDataIO, the only ones you can build a paid product on. "Ingest everything" isn't real; the good data is contracted and metered.

Grounding & anti-hallucination

A product LLM should answer only from retrieved data, show its source, and say "not in the data" instead of guessing. The way you prove it: an eval that scores grounding, honesty, and refusal, not vibes. Measuring quality is half the product.

Engagement → revenue, the loop

More visits and stickier on-site time feed more ad impressions and higher-intent monetization. The job of an on-site AI feature (assistant, game, FAQ) is to keep fans on the page asking instead of bouncing to a search engine.

Monetization Copilot

Reads a piece of content and returns a monetization plan with a compliance verdict.

In: article / page · Out: ad + native plan, betting layer, compliance flags

Compliance Engine

Decides whether a betting/DFS unit can run against a given piece of content, and why.

In: content + proposed unit · Out: allow / restrict / block + reason

Recap-to-Social

Turns a recap into a platform-ready social pack in the team's voice.

In: recap · Out: X / IG / TikTok / push copy

Grounded Sports Q&A

Answers fan questions from live data, cites sources, refuses out-of-scope. (Live: the LLM assistant.)

In: question · Out: grounded answer + source

Sports Trivia Generator

Produces verifiable multiple-choice questions from a stats dataset.

In: stats data · Out: grounded MCQs

Yield & Pricing Assistant

Suggests placements and price ranges from historical performance, honest ranges only.

In: inventory + history · Out: placement + price guidance

Anthropic Claude API

The model layer behind every skill, reasoning, generation, grounding.

In use

ESPN public endpoints

Live NFL scoreboard, standings, schedules, news, prototype data.

In use (prototype)

Webflow / WordPress

Publisher CMS embed, drop the assistant or a game onto any page.

Available

Sportradar / SportsDataIO

Licensed feeds for accuracy, player stats, and commercial rights.

Planned

Google Ad Manager

Serve the native and video units the Monetization Copilot recommends.

Planned

X / Meta / TikTok

Push the social packs out to distribution.

Planned