The quality of an AI coaching session is largely decided by the prompt and the data you hand it. These guides are working notes on how to use ChatGPT and Claude as a sharp, fast research assistant for solo queue improvement.
Read the post-game scoreboard the way a coach would. Role-aware benchmarks, what each number actually measures, and how to combine them before handing the JSON to an AI.
A field-by-field walkthrough of the JSON LoL2LLM hands to ChatGPT or Claude. Knowing what you're sending makes your prompts — and the AI's answers — much sharper.
When to use which of LoL2LLM's five personas, plus the one-line additions that turn a generic answer into a sharp one.
Below Plat, the metrics that move with rank aren't KDA or win rate — they're five much quieter numbers. Improve them and the climb compounds.
Jungle gives the least real-time feedback of any role. The fix is a post-game review framework grounded in your own export — here's the one I use.
An ADC's real job is sustained damage in fights, not staying alive. Four metrics that catch the "safe but useless" pattern KDA can't see.
How vision score is calculated, the per-role targets that actually matter, and the three timing rules that determine where wards go.
ChatGPT and Claude are excellent at LoL analysis — but they fail in five predictable ways. Here's how to neutralize each with a single prompt addition.