> ## Documentation Index
> Fetch the complete documentation index at: https://docs.performax.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Assistant

> Pattern descriptions drawn from your own trading history.

The AI Assistant describes patterns in your own Performax data. It is a review surface, not a trading system.

**Not financial advice.** AI output is not investment, financial, legal, tax, or trading advice. It does not predict markets, generate signals, or tell you what to buy or sell.

<img src="https://mintcdn.com/performax/hkpg9NSlzXmyRSJf/images/ai-demo.png?fit=max&auto=format&n=hkpg9NSlzXmyRSJf&q=85&s=906011f61992ecb2636f5846a4375daf" alt="Performax demo AI Assistant showing a morning briefing, risk observations, focus areas, and chat panel" width="1440" height="1050" data-path="images/ai-demo.png" />

*Demo account. Verify AI text against Journal and trade details before using it in review.*

## What it can use

Depending on available data, the AI can reference:

* closed trades;
* account and period filters;
* strategy/playbook assignments;
* Journal notes and tags;
* behavioral patterns such as overtrading or revenge-trading indicators;
* risk settings and Coach violations;
* saved insights and prior conversations where available.

If the account has little data, the output should be treated as low-confidence review text.

## Daily briefing

The daily briefing summarizes recent account activity:

* prior-day performance;
* notable changes in trade frequency, position size, or realized result;
* risk observations based on configured thresholds;
* focus areas for review.

It should not be read as a trading plan for the day.

## Chat

Use chat to ask questions about your own data, for example:

* `How did my BTC trades perform this month?`
* `Which strategies had the worst drawdown?`
* `Show patterns in my losing trades last week.`
* `Compare my first trade of the day with the rest of the session.`

Good questions specify account, period, symbol, strategy, or behavior. Vague questions can produce vague answers.

## Pattern detection

The AI can describe patterns such as:

* higher loss rate after a sequence of losses;
* increased position size after wins or losses;
* session-specific performance differences;
* symbol concentration;
* strategy performance drift;
* repeated mistakes in Journal tags.

Pattern text should include the sample it is based on when possible. Small samples are not reliable proof of an edge.

## Tilt alerts

Tilt detection watches for behavior that can indicate emotional trading, such as:

* higher trade frequency after losses;
* position size increasing after a loss;
* trading outside normal hours;
* repeated rule violations.

These alerts are review prompts. They do not pause exchange trading and they do not prove intent.

## Saved and pinned insights

Save an insight for later review. Saved insights are for preparation and journaling. They should not be treated as rules unless you explicitly convert them into a strategy checklist or Coach rule.

## Limits

AI output can be incomplete or wrong when:

* trade history is missing;
* filters exclude relevant trades;
* cashflows or balances are incomplete;
* notes contain ambiguous language;
* sample size is too small;
* the model summarizes a pattern too confidently.

Verify important claims against the Journal, Analytics, and raw trade details.