This page is for traders evaluating both apps. The structure: what Kyra is, what Profit AI is, feature-by-feature comparison, and a closing on which app fits which trader.
What Kyra is
Kyra is a privacy-first trading journal for iOS. Trades are stored on your device. Pattern detection runs on-device and waits for sample size before claiming. Every pattern surfaced shows its sample size and an uncertainty range — the four tiers (Tracking, Hint, Signal, Proven) sharpen as your history grows so claims stay honest about how much evidence supports them.
Kyra does not analyze charts, does not generate trade signals, does not emit per-trade verdicts, does not connect to brokers via OAuth, and does not run an AI chatbot. The architectural commitments live on Manifesto; the architectural reasoning is on Why on-device and The math behind Kyra.
Pricing is $4.99/mo or $39.99/yr after a 7-day trial. One free tier, two paid plans, no weekly tier.
Primary buyer: a retail self-executor — a trader with a $10K–$100K account who has a strategy and wants to understand why they cannot always execute it.
What Profit AI is
Profit AI is a cross-platform (iOS, iPad, macOS, Vision) AI trading assistant. The product centers on a vision-LLM that accepts chart input three ways (screenshot upload, TradingView image URL, or live camera capture), extracts indicator readings (RSI, MACD cross zones, moving averages) and chart patterns (double tops, head & shoulders, breakout flags), and emits a trade plan: Entry Zone, Stop Loss, Take Profit, plus a confidence rating.
Around the chart-analysis hero, the app offers:
- A Strategy Builder that lets the user define rules per style (Day, Swing, Scalping, Long-Term), then scores each uploaded chart against the rules using a four-tier label: Ideal / Good / Not Ideal / Poor. The label is a per-trade verdict.
- A trade journal with broker OAuth auto-sync (MT4, MT5, FTMO, TopStep, Tradovate, Interactive Brokers) and a P&L Calendar heatmap (e.g., a green "+$9,100" day-block) with win/loss streak counters.
- A Trader Score (0–100 single number) combining win rate, profit factor, risk discipline, consistency, strategy adherence, and risk-reward ratio with hardcoded weights.
- A Leak Discovery surface that highlights behavioral mistakes flagged by the LLM (e.g., widening stop losses, over-leveraging into NY market opens).
- An AI Mentor chat (24/7 conversational interface).
- A 5-module Academy in Duolingo-style gamified format.
- Pine Script translation for converting saved strategies into TradingView backtests.
Profit AI runs server-side LLM inference (their models plus "trusted third-party model providers acting as our processors"). Account creation is required (email + password + SSO). Trades, chart screenshots, journal notes, and AI-mentor conversations all transit Profit AI's servers. The App Store privacy label discloses an active cross-app tracking SDK.
Pricing spans multiple SKUs including weekly, monthly, and annual tiers, with a short trial. The specific price points and trial terms are visible on the App Store listing, as is current user sentiment in the reviews.
Primary buyer (per Profit AI's broker-page documentation, which names FTMO Phase 1, Funded accounts, and daily-loss-limit alerts as central use cases): a prop-firm trader running funded-challenge or funded-account workflows.
Feature comparison
| Dimension | Kyra | Profit AI |
|---|---|---|
| Primary persona | Retail self-executor | Prop-firm trader (FTMO, TopStep, IBKR) |
| Storage | On-device SwiftData + optional iCloud user-private sync | Cloud servers |
| Account required | No | Yes (email + password + SSO) |
| Cross-app tracking SDK | None | Active (per App Store privacy label) |
| Pattern detection approach | Sample-size-gated statistical inference, on-device | Vision-LLM outputs + LLM-over-journal-data, server-side |
| Sample size shown per claim | Yes — every pattern card shows n and an uncertainty range | Not surfaced |
| Per-trade trade-plan output | None (not a recommendation engine) | Entry Zone / Stop Loss / Take Profit + confidence rating |
| Per-trade alignment verdict | None | Ideal / Good / Not Ideal / Poor (vs saved strategy) |
| Chart screenshot upload | No | Required for hero feature (screenshot, URL, or camera) |
| Indicator extraction (RSI, MACD, MAs) | No (Kyra reads trades, not charts) | Yes |
| Pattern shape detection (double top, H&S, flag) | No | Yes (pixel-coordinate scanning) |
| Broker auto-sync via OAuth | No | Yes (MT4 / MT5 / FTMO / TopStep / Tradovate / IBKR) |
| Trade entry | Manual entry | Manual entry and broker import |
| P&L calendar with heatmap | No (sober list view in History tab) | Yes |
| Win/loss streak counter | No (deliberately — see Manifesto §4) | Yes |
| Single discipline score | None | Trader Score 0–100, hardcoded weights, opaque factor formulas |
| Behavioral pattern output | Pattern engine with explicit sample size and uncertainty range, per pattern | Leak Discovery via LLM, opaque mechanism |
| AI chatbot mentor | No | Yes (24/7) |
| In-app education | No | 5-module Academy, Duolingo-style |
| Pine Script translation | No | Yes |
| Prop-firm sub-account mapping | No | Yes (Phase 1 / Funded vocabulary in UX) |
| Pricing | $4.99/mo, $39.99/yr — two SKUs | 5 IAPs incl. dual weekly tiers, dual annual tiers |
| Trial | 7 days | 3 days |
| Platforms | iOS | iOS / iPad / macOS / Vision |
| Outbound network for core function | None (sync only) | Required for every analysis feature |
The table is neutral. Cells are factual. Both apps work; they work on different problems for different traders.
On per-chart verdicts
The most architecturally distinct line in the table is "Per-trade alignment verdict." Profit AI accepts a chart upload, compares against the user's saved strategy rules, and emits a label — Ideal, Good, Not Ideal, or Poor. The label is a verdict on a specific possible trade.
A verdict on a specific trade is, definitionally, a recommendation. Chart-analysis-with-trade-plan apps as a category tend to ship per-trade verdicts in the feature list while disclaiming "not financial advice" in the footer. Both framings can be true simultaneously only in form, not in function — the user who pays for the verdict is paying for the recommendation.
Kyra does not give per-trade verdicts. Kyra's output is statistical observation about the trader's past trades — never a verdict on the next one. The product is the trader's history, structured. Not an AI's opinion of an opportunity.
The full reasoning on the recommendation-engine refusal lives on Manifesto (Section 2: Not a recommendation engine).
On the Trader Score
Profit AI's Trader Score (0–100) is a composite of several discretionary factors with weights chosen by the developer. The underlying formula isn't documented publicly, and no sample-size gating is described.
Kyra does not display a single discipline score. It presents individual patterns with sample sizes and uncertainty ranges, so you can assess each claim without collapsing your behavior into one opaque number.
On Leak Discovery vs Kyra's pattern engine
Profit AI's Leak Discovery surface flags behavioral mistakes — widening stop losses, over-leveraging into NY market opens — by passing journal data through an LLM and surfacing what the model identifies as problematic. The named examples are reasonable; the mechanism is opaque. A trader cannot ask "how many trades was this based on" or "how much confidence is there in this gap" because the model does not surface its working.
Kyra's pattern engine answers the same question differently. The engine looks across the dimensions a trader is already capturing — emotion, setup, execution rating, time-of-day — and waits until enough evidence accumulates before promoting a candidate into a claim. When a pattern reaches the Hint, Signal, or Proven tier, it shows the sample size and an uncertainty range so the trader can see exactly how much evidence supports it.
There is also a copy-frame difference worth naming. Profit AI's Leak Discovery surface organizes output into three columns: Strengths, Leaks, and Fixes. The Fixes column is prescriptive — it tells the trader what to do. Kyra deliberately does not have a Fixes column. The engine surfaces observations; what the trader does with them is the trader's call. Per the Manifesto commitment to "not a coach," Kyra never writes a sentence in the imperative mood about the next trade.
The output shapes are similar. The mechanism honesty is not, and the coach-versus-mirror framing is not.
On privacy
Apple's App Privacy nutrition labels are the easiest verification. Compare the labels on each app's App Store listing.
Profit AI's label discloses: data used to track you (usage data), data linked to your identity (account info, purchases, billing), data not linked to your identity (device ID, product interaction, crash data) — used for third-party advertising plus analytics plus functionality.
Kyra's label discloses: device ID and product interaction (used for analytics only), crash data (used for app functionality). No data linked to user identity. No data used to track across apps and websites of other companies. No trade content on any server Kyra operates.
The structural difference is whether trade content transits a server the app's developer operates. Profit AI: yes, by architecture. Kyra: no, by architecture. Neither app is lying about its label; the labels accurately describe each architecture. The trader's choice is which architecture they want their journal to use.
Who should pick which
The trade-off is real and the personas are different. Pick Profit AI if any of the following matters more to you than on-device storage or visible statistical evidence:
- You are running an FTMO, TopStep, Tradovate, or IBKR account and want the broker sync built around prop-firm workflows (Phase 1 evaluation, Funded account, daily-loss-limit alerts)
- You want an AI to analyze a chart screenshot and emit a trade plan
- You want a per-chart verdict (Ideal / Good / Not Ideal / Poor) against your saved strategy
- You want broker auto-sync via OAuth
- You want a 24/7 chat mentor
- You want gamified in-app education
- You want a P&L calendar heatmap and win/loss streak tracking
- You want Pine Script translation for TradingView backtesting
- You want the same product on iOS, iPad, macOS, and Vision Pro simultaneously
- You want a single 0–100 score regardless of sample size
Pick Kyra if any of the following matters more:
- You are a retail self-executor trading your own equities or options account
- You want your trade data on your device and nowhere else
- You want a journal with no account creation and no email collection
- You want pattern claims with explicit sample sizes and uncertainty ranges shown on every card
- You want a journal that waits for evidence before claiming a pattern, rather than emitting confident-sounding sentences regardless of sample size
- You want predictable pricing with no weekly tiers and no A/B price testing
- You want the journal to be a mirror, not a coach
- You want a sober UI without heatmaps, streak counters, badges, or gamification
Neither set of preferences is wrong. The two apps are aimed at different buyers with different problems.
What this page does not say
This page is a comparison. It is not a put-down. Profit AI is competently built and useful for the trader it is built for. The architectural choices on each side are deliberate; neither is sloppy.
What this page does not say:
- That Profit AI's chart analysis is inaccurate — the trader can evaluate that directly
- That Profit AI's broker auto-sync is poorly implemented — it is widely used in the prop-firm community
- That Profit AI's pricing is exploitative — the weekly tiers are unusual but legal and publicly visible
- That Profit AI's users are wrong — the 2,100+ reviews suggest the product fits a real audience
What this page does say:
- The architectures are opposites (cloud + LLM + accounts vs on-device + classical statistics + no accounts)
- The trade-offs are real and visible
- The privacy nutrition labels are auditable and tell the architectural story honestly
- The statistical-vs-LLM choice fits different traders
- The retail-self-executor vs prop-firm-trader persona split is the largest determining factor — pick the app aimed at the buyer you are
If the on-device + statistical-rigor + retail-self-executor side of the trade-off is what you wanted from a journal, Kyra is the app. If the AI-mentor + chart-analysis + prop-firm-workflow side is what you wanted, Profit AI is the app. The category is finally big enough to support both kinds of answer.
Download Kyra on the App Store · Why on-device · The math behind Kyra · The full security model · Compare to TradeZella · Compare to a spreadsheet
