PRODUCT FACTS · 2026-10-21
NoahAI: current capabilities and public release
Financial AI that helps people understand decisions and remain in control of their accounts. Distinguish the capabilities in public release 3.9.2.20 from real external-environment validation that remains pending.
Free distribution since 3.9.2.0 · Selected paid plans
Current public Windows release: v3.9.2.20 · 2026-10-21
Windows updater: 3.9.220
External acceptance: pending; see the current release manifest. Publication does not establish actual installation, automatic updates, venue/OCX, rollback, real microphones or 24/72-hour PAPER acceptance. Windows files are unsigned.
Official release and Windows assets · Release manifest · Sign in for member downloads
Current public Windows release and release history
Is NoahAI free?
Free distribution began with 3.9.2.0, alongside selected paid plans. A free download does not make every feature, venue or external AI API free. Membership rules distinguish domestic spot access, overseas exchange referral/UID approval and paid asset coverage. The free public Strategy Hub test is separate from paid Client plans.
What does autonomous financial technology do?
The engine on the user's PC connects market observation, candidate selection, regime/cost/risk checks, execution or HOLD, and outcome records. It operates only within authorized accounts, venues and modes. An external LLM response cannot grant trading permission. Web remote support does not move the trading engine to the cloud. Neither profits nor uninterrupted, error-free operation are guaranteed.
Why use Strategy Studio?
Connect source documents, natural language and TradingView/Pine ideas to executable rules, clarify missing conditions, and retain evidence and results for each approved version. Levels 1–2 focus on accessible explanations and key values; 3–4 on detailed rules and operating policies; Level 5 on cost sensitivity and validation evidence. Higher levels do not remove permissions or risk protection. Not every Pine script or arbitrary program is supported.
Is backtest approval required? What are PAPER and LIVE?
Historical replay/backtesting is optional. Valid executable rules, user approval and applicable execution checks allow PAPER forward validation without passing a backtest. PAPER simulates trades using current market data; LIVE places real orders in the user's account. PAPER results are neither confirmed fills nor proof of future returns. LIVE needs separate permissions, account readiness and risk checks.
Do 40 saved strategies run at once?
No. Per-venue PAPER strategy evaluation has up to 10 slots for general members and 30 for Premium. The applied strategy also uses a slot. Eligible, evaluated and waiting strategies are distinguished; saved-strategy and position limits are separate. Waiting is not execution or automatic rotation.
What is available in everyday finance?
Cumulative public v3.9.2.13 capabilities include user-entered financial context, planning, insurance discovery/comparison and encrypted My Money Plan storage with monthly actual-record reviews. Initial insurance discovery covers four companies and seven guidance materials, not a full-market live feed or binding personal quotes. Unknown rates, costs, coverage and after-tax amounts remain unconfirmed. It does not establish automatic bank transfers, enrollment, cancellation, tax filing or optimal personal tax outcomes.
What is the current public Windows version?
The public release is v3.9.2.20. The Windows installer adds consent/withdrawal history and client version records. Collection is off by default and previous consent requires reconfirmation. Withdrawal clears unsent activity; offline withdrawal notifications are sent first on reconnect. Real-user counts, return/retention and PAPER/LIVE performance remain unmeasured. Test, internal and unclassified accounts are excluded. Real activity requires reviewed classification evidence and registered consent. Activity before classification approval is excluded; missing historical consent or version is never inferred. Existing user/trade data and v3.9.2.19 PAPER exit recovery are preserved.
How do publication and real-user KPIs differ?
The public release is v3.9.2.20. The Windows installer adds consent/withdrawal history and client version records. Collection is off by default and previous consent requires reconfirmation. Withdrawal clears unsent activity; offline withdrawal notifications are sent first on reconnect. Real-user counts, return/retention and PAPER/LIVE performance remain unmeasured. Test, internal and unclassified accounts are excluded. Real activity requires reviewed classification evidence and registered consent. Activity before classification approval is excluded; missing historical consent or version is never inferred. Existing user/trade data and v3.9.2.19 PAPER exit recovery are preserved.
How are KPIs defined?
Active users are evidence-reviewed, consented external-use accounts with explicit activity in a KST window after classification approval. Test, internal and unclassified accounts, status checks and background-only analysis are excluded. Week-over-week return is distinct from signup-cohort D7 retention. Missing coverage or denominators remain unmeasured, not zero. New collection contains pseudonymous account keys, action types, times, PAPER/LIVE modes, consent/withdrawal history, client versions and random installation identifiers; it excludes names, email, exchange keys and strategy text.