AI Transparency
Last updated: June 2025
Our Commitment
Generative AI is everywhere in consumer software, and not all of it is disclosed. We believe you deserve to know exactly where AI is and is not involved in what SnapPick shows you — especially when you are making decisions about your game based on our output.
This page is a complete, specific account of AI use across SnapPick as of the date above. We update it whenever our use of AI changes materially.
Where AI Is NOT Used
- Recommendation outputs. Brawler picks, deck recommendations, upgrade priorities, and tier list placements are computed by a deterministic, rule-based scoring engine. The engine does not use a language model, a neural network, or any probabilistic AI. Given the same input data, it always produces the same output.
- Game data. Win rates, pick rates, card statistics, brawler attributes, map data, and event rotations come directly from the Supercell API and verified third-party sources. We do not generate, estimate, or hallucinate any game numbers.
- Player data. Your card collection, battle history, and brawler roster are fetched directly from the Supercell API using your player tag. AI is not involved in interpreting or transforming this data.
Where AI IS Used
| Area | How AI is used | Human review |
|---|---|---|
| Strategy guide drafting | AI writing tools (e.g. Claude) may be used as a drafting aid for structure, phrasing, and language quality when producing long-form guides. | Every factual claim, score, and strategic recommendation is verified by the editor before publication. AI-assisted guides are labelled as such. |
| Translation assistance | Machine translation (via AI tools) is used to produce initial drafts for non-English content. This reduces cost and time while keeping the site multilingual. | Non-English content is reviewed before being surfaced to search engines. |
| Development tooling | AI coding assistants help write and review SnapPick's source code. This is standard software development practice and does not affect content or data. | All code is reviewed and tested by the developer (Diego García / digsadev) before deployment. |
Why Our Recommendations Are Not AI-Generated
Generative AI models can produce confident-sounding but incorrect game advice, especially for fast-moving metas where their training data is stale. A language model cannot know what happened in last week's balance patch.
Our rule-based engine, by contrast, is deterministic and auditable. When you see a recommendation, the scoring factors are shown to you (win rate contribution, synergy bonus, ownership fit). You can disagree with the weights — but you cannot be silently misled by a hallucination.
Questions or Concerns
If you believe AI-generated content has appeared on SnapPick without proper disclosure, or if you have questions about any specific page or feature, contact us at snappickapp@gmail.com.