The analyst

The Noxint chat is an AI market analyst for crypto and US stocks, and it runs on one rule: the model’s memory is not a source. Every answer is grounded in data pulled live at ask-time. If the data is not there, the analyst says so instead of improvising a number. Chat is also the live proof of the engine. The same data layer that serves agents over the API answers you here, under the same freshness clocks and the same refusal rules.

How it works

When your question needs a number, the analyst calls one of 23 live data tools spanning crypto, equities, market context, and web search. You watch it work through plain-language activity labels: “Pulling the live tape”, “Mapping liquidation clusters”, “Computing indicators · BTC · 4H”, “Tracking Form 4 activity”. The final answer is built from those reads, nothing else. The tools reach across the whole picture:

Live crypto prices with cross-venue consensus, funding, open interest, taker flow, liquidations, order-book depth, indicators, and market structure
On-chain token lookup, trending flow, and contract-safety screening
US equity financials exactly as filed, plus valuation, growth, price history, and the filing stream: insider (Form 4) trades and institutional (13F) ownership included
Market context: news flow, prediction-market odds, the X/CT social read, and the macro regime (dollar, yields, equity vol, BTC dominance)
Web search, for narratives the data tools do not carry

What the analyst reads

When the analyst pulls the Bitcoin situation, the engine hands it a deterministic, rule-versioned assessment like this one, captured live on 2026-07-16:

{
  "assessment": {
    "headline_regime": { "state": "ranging", "rule": "headline_regime@1", "adx_14": 28.43 },
    "positioning": { "state": "mild_long", "funding_band": "leaning_long", "oi_trend": "unwinding", "rule": "funding_oi@1" },
    "structure": { "trend_state": "neutral", "invalidation": { "price_usd": 64411.8, "distance_pct": 0.54 }, "rule": "structure_invalidation@1" },
    "flow": { "state": "bid_dominant", "lead": "perp_led", "confirms_price": false, "rule": "taker_flow@1" },
    "cascade_risk": { "level": "low", "nearest_shelf": { "side": "short", "price_usd": 64327.42, "mass_usd": 194099234 }, "rule": "shelf_cascade@1" },
    "risk_flags": ["flow_divergence", "single_venue_scope", "stale_inputs"],
    "confidence": { "level": "medium", "evidence": { "proved": 18, "refused": 0 } }
  }
}

Positioning, structure, flow, and cascade risk, each state stamped with the rule version that produced it, plus an evidence ledger of what was proved and what was refused. The analyst’s job is to reason over facts of this quality and answer your actual question.

Personas

Three analysts, one data layer. Quant is the default: balanced, leads with synthesis. Degen hunts momentum and setups and talks like it. Contrarian measures crowding and looks for the fade. All three run the same tools and receive the same payloads; only the register changes. Nobody gets to make up a different tape.

Live reasoning trace

A collapsible panel streams the analyst’s thinking while the answer generates. It exists only while the answer is live: filtered out before storage, gone after a reload. Details in privacy.

Working with answers

Attach up to three images per message (JPEG, PNG, WebP, GIF, or AVIF, 5 MB each): a chart screenshot is a valid question.
Edit any question you asked to re-run it against fresh data.
Follow-up chips propose the next question after each answer, and your threads live in a paginated sidebar history.

A Telegram surface is live: the same analyst and the same meter, in DMs.

Getting started

Chat

Token

Data

How it works

What the analyst reads

Personas

Live reasoning trace

Working with answers

​How it works

​What the analyst reads

​Personas

​Live reasoning trace

​Working with answers

How it works

What the analyst reads

Personas

Live reasoning trace

Working with answers