RangeSharp
RangeSharp
AI analyzes your training mistakes and finds conceptual leaks. Not just what you got wrong, but why.
You finish a training session. 73% accuracy. You missed 12 hands. The tool shows you which hands: ATs, K9s, 44, QJo. You look at them. You see the right answer. You tell yourself you'll remember next time. Then you close the tab and forget which hands you missed by tomorrow.
That's what most poker trainers give you. Accuracy. A mistake list. No explanation of why you're making those mistakes. No pattern. No "here's what you're doing wrong at a conceptual level."
FreeRangeLab's AI Training Debrief is the missing piece.
After a session, you hit "Get AI Debrief." The system sends your session data to an LLM: every hand you saw, every correct answer, every mistake, the positions, the action lines. The AI looks for patterns. Not "you folded ATs when you should have raised." That's just a fact. The AI looks for stuff like: "you fold too many suited broadway hands when facing 3-bets from the blinds" or "you overvalue small pocket pairs in late position vs early opens."
The output is a short analysis. A few paragraphs. Actionable. It points at a leak, explains why it might matter, and suggests what to study. You get the "why," not just the "what."
Here's a realistic example. I ran a 50-hand session on BB Defense vs BTN. Missed 14. Got the debrief:
Session: BB Defense vs BTN Open. 72% accuracy. 14 mistakes.
Your errors cluster around two areas:
Suited connectors and suited aces: You folded 65s, 76s, A5s, and A6s when the correct action was call. These hands have good playability vs a BTN open. They realize equity well in position, and the BTN's range is wide enough that you're not dominated. You might be over-folding "speculative" hands. Consider drilling this hand class specifically.
Marginal Broadway hands: You called with KJo and QJo when the solution folds. These hands are tricky. They look strong, but they're often dominated by the BTN's value range and don't have enough equity vs their bluffs. Your tendency to call suggests you might be over-defending with hands that look good but play badly. The fix: treat KJo and QJo as folds in this spot until you've internalized the solver output.
Recommended focus: Suited connectors and suited aces in BB vs BTN. Run a Recall mode session on that subset. Then re-test.
That's the kind of thing you get. Not "you got 65s wrong." A pattern. A reason. A next step.
Finish a training session. Decision mode or Recall mode.
On the session summary screen, click "Get AI Debrief."
Wait a few seconds. The analysis appears.
Read it. Note the patterns.
Go fix the leak. Open the spot, drill the weak hand class, or run a focused Recall session on the hands you're messing up.
The debrief doesn't replace studying. It tells you where to study. You still have to put in the work. But now you know what to work on.
Free: 1 debrief per day. Enough to get a taste. Pro: 10 per month. Elite: 30 per month. If you're training daily, Pro or Elite makes sense. One debrief per session is usually enough. You don't need one per hand.
Accuracy numbers are useful. They tell you you're at 73% instead of 90%. They don't tell you that your problem is suited connectors, or that you're over-calling with KJo. Without that, you're shooting in the dark. You review your mistakes, you see ATs, K9s, 44. You don't see "I'm folding too many hands that flop well" or "I'm calling too many hands that get dominated."
The debrief connects the dots. It turns a list of wrong hands into a conceptual leak. Then you can target that leak. One session on suited connectors in BB vs BTN beats ten sessions of random drills where you hope you're improving.
The model gets: hand dealt, your answer, correct answer, position, action line, and any notes on the spot. It doesn't get your identity or your history across sessions. It's per-session. The analysis is based on that session's mistakes, not your entire training history. That keeps it focused and private.
The AI is a pattern detector, not a solver. It can miss subtle patterns or occasionally misidentify one. If the debrief says "you over-fold suited aces" and you're pretty sure you're not, double-check with your actual mistake list. The debrief is a guide, not gospel. Use it to direct your study. Verify when something feels off.
Also, the quality of the analysis depends on how many mistakes you made. A session with 3 mistakes might not yield a strong pattern. A session with 15 might. Run longer sessions when you want a useful debrief.
Most training tools answer "what did I get wrong?" They show you the hands. They don't answer "why am I getting these wrong?" That second question is where real improvement lives. The AI Training Debrief is built to answer it. Not with generic advice. With patterns specific to your session. Your mistakes. Your leaks.
Without the why, you're stuck in a loop. Miss ATs. See the answer. Tell yourself you'll remember. Miss it again in three days. Repeat. The debrief breaks the loop. It says "you're folding suited broadways when you shouldn't" or "you're over-calling with dominated hands." Now you have a target. Drill suited broadways. Drill the fold range. The next session, the pattern might shift. The debrief adapts. Each session gets its own analysis.
The debrief lives in the session summary. You finish. You see accuracy, EV loss, streak. Below that, "Get AI Debrief." Click it. Read. Then use the study recommendations. The dashboard has a "Weak spots" section. The debrief might say "focus on suited connectors in BB vs BTN." That spot will show up in Weak spots. You drill it. The two systems reinforce each other. The debrief tells you what to study. The analytics show you where you're weak. Same information, different angles.
The debrief is included in Free (1/day), Pro (10/month), and Elite (30/month). No extra add-on. It's part of the training flow. Finish a session, get the analysis, move on. Simple. There's no separate "AI coach" product to buy. It's built into the session summary. You use it when you want it. I skip it after strong sessions. No point asking "why did I mess up?" when I scored 92%. I use it when something felt off. When I knew I was guessing. When the accuracy number made me grimace. That's when the debrief earns its place.
Go fix them.
It's not a replacement for reviewing your mistakes manually. You still need to look at each hand. See what you picked. See the correct answer. Feel the gap. The debrief summarizes. It doesn't replace the moment of "oh, I folded that? Why?" That moment is where learning happens. The debrief tells you which moments to pay extra attention to. You should still look at the hand list. See the correct answer. Think about why you got it wrong. The debrief gives you a lens. It doesn't replace the work. It tells you where to focus.
It's also not a strategy coach. It won't tell you to play more aggressively or tighten up. It analyzes patterns in your mistakes within a single session. It doesn't have access to your full history or your win rate. It's a study aid, not a mental game coach.
One debrief after one bad session might not reveal much. But if you run debriefs regularly, you'll start seeing recurring themes. "You over-fold suited connectors" might show up in BB vs BTN, BB vs CO, and BB vs HJ. That's not a spot leak. That's a hand-class leak. The debrief surfaces it. You drill suited connectors across spots. The leak shrinks.
I run a debrief maybe twice a week. After sessions where I felt lost or scored below 75%. I don't need one after every session. Just when I want to understand what went wrong. The Pro plan gives 10 per month. That's enough for 2-3 per week. Elite gives 30. If you're training multiple spots daily and want a debrief after each weak session, Elite makes sense. Most people will be fine with Pro.
One more thing: the debrief is only as good as your session data. If you ran 10 hands and missed 2, the AI has almost nothing to work with. Run at least 30-50 hands when you want a useful debrief. More mistakes means more pattern detection. A 20-hand session with 90% accuracy won't yield much. A 50-hand session with 70% accuracy will.
Master your preflop ranges.